System

Abstract

A system is provided.SOLUTION: The system includes means for collecting data such as a payment history, a communication history, and a financial situation of a debtor, means for analyzing the collected data and generating an optimal reminder plan for each debtor, means for automatically executing a reminder activity for the debtor based on the generated reminder plan, means for searching for and tracking the latest contact information of the debtor who cannot be contacted, means for generating a document for supporting a necessary legal procedure, and means for collecting a result of the reminder activity and updating a model based on an analysis result.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The technology of the present disclosure relates to a system. [Background technology]

[0002] Patent document 1 discloses a persona chatbot control method performed by at least one processor, the method including the steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to a description of the chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance in response to the user utterance. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Publication No. 2022-180282 Summary of the Invention [Problem to be solved by the invention]

[0004] Currently, the debt collection process is inefficient, requiring a lot of manual work and time, placing a heavy psychological burden on debt collection personnel. Furthermore, if a debtor becomes unreachable, tracking them down is cumbersome and requires a lot of human resources. Therefore, the present invention aims to automate and optimize the debt collection process, thereby improving the efficiency and speed of debt collection work and reducing the burden on debt collection personnel. [Means for solving the problem]

[0005] The present invention provides a means for collecting and analyzing data such as debtors' payment history, communication history, and financial status. Furthermore, the system incorporates a means for analyzing the collected data and generating an optimal collection plan for each debtor, enabling effective and efficient debt collection activities. It also includes a means for automatically carrying out collection activities based on the generated collection plan and a means for tracking the latest contact information of debtors who have become unreachable. This streamlines the debt collection process and saves human resources and time. Furthermore, a means for generating documents to support necessary legal procedures is provided, and the results of collection activities are collected and the model is updated to improve the accuracy of future collection plans.

[0006] "Debtor" means an individual or legal entity that owes a particular debt.

[0007] "Payment history" refers to historical information about payments made by a debtor in the past, and is data including the date and time of payment, the amount, and the payment method.

[0008] "Communication history" refers to records of communications such as telephone calls and emails between creditors and debtors, and is data that includes information such as the date and time of communications, content, and response status.

[0009] "Financial situation" refers to data that indicates information about the debtor's income, deposit balance, loans, and other financial assets and liabilities.

[0010] "Data collection means" refers to a device or system for collecting data such as payment history, communication history, and financial status in a specific manner.

[0011] "Data analysis means" refers to a device or system that analyzes collected data and extracts specific patterns or trends.

[0012] A "collection plan" refers to a plan that includes the optimal collection methods and timing for a debtor, and is an individually customized plan.

[0013] "Collection activities" refers to actions taken to request payment from debtors, and includes means such as telephone, email, and written notices.

[0014] An "uncontactable person" refers to a debtor who, for some reason, the creditor is unable to contact.

[0015] "Tracking means" refers to a device or system used to identify the latest contact information and address of an unreachable debtor.

[0016] "Legal procedure documents" refer to documents that support the legal procedures necessary for debt collection, including mediation procedure documents and bankruptcy procedure documents.

[0017] "Model" or "Generative AI Model" refers to an artificial intelligence model that uses machine learning algorithms to generate an optimal reminder plan based on the analysis results.

[0018] "Updating" refers to the process of retraining the generative AI model based on the results of reminder activities to improve the accuracy of future reminder plans. [Brief explanation of the drawings]

[0019] [Figure 1] 1 is a conceptual diagram showing an example of the configuration of a data processing system according to a first embodiment. [Figure 2] 1 is a conceptual diagram showing an example of main functions of a data processing device and a smart device according to a first embodiment. [Figure 3] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a second embodiment. [Figure 4] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and smart glasses according to a second embodiment. [Figure 5] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a third embodiment. [Figure 6] FIG. 11 is a conceptual diagram showing an example of main functions of a data processing device and a headset-type terminal according to a third embodiment. [Figure 7] FIG. 10 is a conceptual diagram showing an example of the configuration of a data processing system according to a fourth embodiment. [Figure 8] FIG. 10 is a conceptual diagram showing an example of main functions of a data processing device and a robot according to a fourth embodiment. [Figure 9] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 10] 1 shows an emotion map onto which multiple emotions are mapped. [Figure 11] FIG. 3 is a sequence diagram showing a processing flow of the data processing system according to the first embodiment. [Figure 12] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 1. [Figure 13] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system according to the second embodiment when an emotion engine is combined. [Figure 14] FIG. 10 is a sequence diagram showing the flow of processing in the data processing system in Application Example 2 when an emotion engine is combined. DETAILED DESCRIPTION OF THE INVENTION

[0020] An example of an embodiment of a system according to the technology of the present disclosure will be described below with reference to the accompanying drawings.

[0021] First, the terms used in the following description will be explained.

[0022] In the following embodiments, a coded processor (hereinafter simply referred to as a "processor") may be a single arithmetic device or a combination of multiple arithmetic devices. Furthermore, a processor may be a single type of arithmetic device or a combination of multiple types of arithmetic devices. Examples of arithmetic devices include a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), a GPGPU (General-Purpose computing on Graphics Processing Units), and an APU (Accelerated Processing Unit).

[0023] In the following embodiments, a coded RAM (Random Access Memory) is a memory in which information is temporarily stored and is used as a working memory by a processor.

[0024] In the following embodiments, the coded storage is one or more non-volatile storage devices that store various programs, various parameters, etc. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), and magnetic tapes.

[0025] In the following embodiments, a communication I / F (Interface) with a symbol is an interface including a communication processor, an antenna, etc. The communication I / F controls communication between multiple computers. Examples of communication standards applied to the communication I / F include wireless communication standards including 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), Bluetooth (registered trademark), etc.

[0026] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." In other words, "A and / or B" means that it may be only A, only B, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" is also applied when three or more things are expressed connected by "and / or."

[0027] [First embodiment]

[0028] FIG. 1 shows an example of the configuration of a data processing system 10 according to the first embodiment.

[0029] 1, a data processing system 10 includes a data processing device 12 and a smart device 14. An example of the data processing device 12 is a server.

[0030] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0031] The smart device 14 includes a computer 36, a reception device 38, an output device 40, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The reception device 38, the output device 40, and the camera 42 are also connected to the bus 52.

[0032] The reception device 38 includes a touch panel 38A, a microphone 38B, and the like, and receives user input. The touch panel 38A detects contact with an indicator (for example, a pen or a finger) to receive user input by the touch of the indicator. The microphone 38B detects the user's voice to receive user input by voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the data indicating the user input.

[0033] The output device 40 includes a display 40A and a speaker 40B, and presents data to the user 20 by outputting the data in a form of expression that the user 20 can perceive (for example, audio and / or text). The display 40A displays visible information such as text and images in accordance with instructions from the processor 46. The speaker 40B outputs audio in accordance with instructions from the processor 46. The camera 42 is a compact digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0034] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 control the exchange of various information between the processor 46 and the processor 28 via the network 54.

[0035] FIG. 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0036] 2, in the data processing device 12, a specific process is performed by the processor 28. A specific processing program 56 is stored in the storage 32. The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific process is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0037] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0038] In the smart device 14, the processor 46 performs the reception output process. The storage 50 stores a reception output program 60. The reception output program 60 is used in conjunction with the specific processing program 56 by the data processing system 10. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0039] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0040] 1. Data Collection

[0041] Subject: Server

[0042] The server first collects data such as the debtor's payment history, communication history, and financial situation. At this stage, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the email communication history is obtained from the email server.

[0043] 2. Data Analysis

[0044] Subject: Server

[0045] The server analyzes the collected data. When analyzing the data, it first preprocesses it to remove noise and fill in missing values. It then uses a machine learning algorithm to predict payment trends and the optimal timing for reminders. Specifically, it inputs past delay patterns into a model to predict the likelihood of the next delay.

[0046] 3. Generate a customized reminder plan

[0047] Subject: Server

[0048] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will create a plan to first remind that debtor by email.

[0049] 4. Execute automatic reminders

[0050] Subject: Server

[0051] The server automatically executes reminder activities based on the generated reminder plan. For example, it automatically generates and sends reminder emails at a specified date and time. If a reminder call is required, it makes a call via an automated calling system and plays a recorded message.

[0052] 5. Tracking Lost Contacts

[0053] Subject: Server

[0054] The server tracks the latest contact information of debtors who have become unreachable, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, allowing lost contact to be restored.

[0055] 6. Legal assistance

[0056] Subject: Server

[0057] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[0058] 7. Feedback and optimization

[0059] Subject: Server

[0060] The server collects the results of collection activities and updates the model. For example, it records the open rate of emails and the response status of phone calls, and reflects this in the next collection plan. This feedback loop improves the accuracy of future collection plans, enabling more efficient debt collection.

[0061] Examples:

[0062] For example, the server retrieves debtor A's payment history for the past year via the bank database and API, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a further reminder phone call is made. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0063] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and enables more efficient and faster debt collection activities.

[0064] The processing flow will be explained below.

[0065] Step 1:

[0066] Subject: Server

[0067] The server collects data such as the debtor's payment history, communication history, and financial situation. This collection process obtains the necessary information using the API of a credit information agency, a bank's transaction database, and a telecommunications company's API. For example, the server sends a query to a bank database to obtain a specific debtor's payment history for the past year. It also obtains communication history (phone and email records) from a telecommunications company's API.

[0068] Step 2:

[0069] Subject: Server

[0070] The server preprocesses the collected data, including cleaning the data (removing noise) and imputing missing values. Specific operations include deleting incomplete data rows and imputing missing values ​​with the mean or median. It also converts data in non-standard formats into a standard format.

[0071] Step 3:

[0072] Subject: Server

[0073] The server analyzes the preprocessed data and extracts payment trends for each debtor. This analysis uses machine learning models to detect debtor delay patterns from past payment history. For example, the server can identify debtors who have had many late payments in the past and calculate the risk of their next payment delay.

[0074] Step 4:

[0075] Subject: Server

[0076] The server generates an optimal reminder plan for each debtor. The generated reminder plan includes the optimal reminder method (phone, email, written) and optimal timing. For example, if a debtor has responded well to emails in the past, the server generates a plan that prioritizes email reminders.

[0077] Step 5:

[0078] Subject: Server

[0079] The server automatically executes reminder activities based on the generated reminder plan. Specifically, it automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder calls via an automated calling system and plays recorded messages.

[0080] Step 6:

[0081] Subject: Server

[0082] The server tracks the latest contact information of debtors who have become unreachable. To do this, it uses a resident registration database or a telecommunications company's API to identify the latest address and contact information. For example, the server accesses a resident registration database to obtain the debtor's latest address information.

[0083] Step 7:

[0084] Subject: Server

[0085] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for delivery to relevant departments and lawyers.

[0086] Step 8:

[0087] Subject: Server

[0088] The server collects the results of the collection activities and records them in a database, for example, by saving information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made.

[0089] Step 9:

[0090] Subject: Server

[0091] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and incorporate new patterns and trends.

[0092] Example 1

[0093] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0094] In conventional debt collection systems, collection plans for each debtor are general and uniform, and collections are not optimally tailored to the circumstances of each debtor. This reduces collection efficiency and makes it difficult to track debtors when they cannot be contacted. Furthermore, preparing documents required for legal procedures is a manual process that consumes time and resources.

[0095] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0096] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for preprocessing the collected data, removing noise, and completing missing values; means for analyzing the data using a machine learning algorithm and predicting the optimal timing and means of collection for each debtor; means for generating a collection plan customized for each debtor based on the prediction; means for automatically carrying out collection activities for debtors based on the generated collection plan; means for searching for and tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of collection activities and updating the model based on the analysis results. This automates the debt collection process, enables optimal collection activities tailored to the circumstances of individual debtors, efficient tracking when debtors cannot be contacted, and enables the rapid and accurate preparation of legal procedure documents.

[0097] A "debtor" is an individual or legal entity that has a financial obligation in the form of a debt, loan, or other form of debt.

[0098] "Payment history" is a record of past payments made by a debtor, including details such as due dates, amounts, and payment methods.

[0099] "Communication history" refers to a record of communication activities such as phone calls, emails, and messages between you and the debtor.

[0100] "Financial status" refers to information that indicates the debtor's current financial condition, and includes data on income, expenses, assets, liabilities, and the like.

[0101] "Data collection means" refers to a method or device by which the server obtains information such as the debtor's payment history, communication history, and financial situation.

[0102] "Data preprocessing" is the process of preparing collected data before analysis, such as removing noise from the data and filling in missing values.

[0103] "Machine learning algorithms" are statistical methods and models for finding patterns and knowledge in data.

[0104] An "analysis means" is a method or device for predicting a debtor's behavioral tendencies and the optimal timing for making a collection request by utilizing collected and preprocessed data.

[0105] A "demand plan" is a plan that determines the optimal means and timing of demands to prevent a debtor from delaying repayments.

[0106] A "customized reminder plan" is a reminder plan that is individually tailored based on each debtor's characteristics and behavior.

[0107] The "automatic prompting activity execution means" is a method or device that allows the server to automatically execute prompting means such as telephone calls or e-mails based on the generated prompting plan.

[0108] A "contact information tracking means" is a method or device for searching for and tracking the latest contact information of a debtor who has become unreachable.

[0109] "Legal procedure support means" refers to a method or device for generating documents related to necessary legal procedures and providing them to relevant departments and experts.

[0110] The "feedback collection means" is a method or device for collecting the results of prompting activities and reflecting them in the next prompting plan.

[0111] A "model updater" is a method or device for retraining an analytical model based on collected feedback data to improve its accuracy.

[0112] The present invention provides a system for automating and efficiently carrying out debt collection activities against specific debtors. A specific embodiment of this system will now be described.

[0113] Hardware and software used

[0114] This system operates in a server-based network environment. The specific hardware and software that supports each process in this system are listed below.

[0115] Server: Collects data, performs analytics, performs remediation activities, tracks contact information, assists with legal proceedings, gathers feedback, and updates the model.

[0116] Example of hardware used: A server equipped with a high-performance CPU and large-capacity RAM

[0117] Examples of software used: MongoDB (database), Pandas (data processing), Scikit-learn (machine learning), Twilio (automated calls), SMTP server (email sending)

[0118] Terminal: A device that interacts with a server and displays or inputs information.

[0119] Examples of hardware used: personal computers, tablets, smartphones

[0120] Specific processing of the system

[0121] Data collection

[0122] The server uses APIs from credit information agencies, banks, and telecommunications companies to collect data such as debtor payment history, communication history, and financial status. For example, the server obtains debtor A's payment history for the past year from the bank API and stores that data in MongoDB.

[0123] Data Preprocessing

[0124] The collected data is first preprocessed. The server uses the Pandas library to remove noise and impute missing values, then formats the collected data into a data frame.

[0125] Data analysis

[0126] Machine learning algorithms are applied to the preprocessed data. The server uses Scikit-learn and models such as random forests to predict the debtor's payment trends and the optimal timing for reminders. For example, it analyzes the debtor's delay patterns and predicts the likelihood of the next delay.

[0127] Generate a customized reminder plan

[0128] Based on the analysis results, the server generates a customized reminder plan for each debtor. This reminder plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will plan to first remind that debtor by email.

[0129] Execute automatic reminders

[0130] Based on the generated reminder plan, the server automatically executes reminder activities. It automatically generates reminder emails at the specified date and time and sends them via the SMTP server. In addition, if a phone reminder is required, it makes an automated call via Twilio's API and plays a pre-recorded message.

[0131] Tracking lost contacts

[0132] For debtors who can no longer be contacted, the server uses the resident registration database or the telecommunications company's API to search for the latest contact information and update the database.

[0133] Legal process assistance

[0134] The server automatically generates documents to support necessary legal procedures, such as mediation and bankruptcy proceedings, and sends them to the relevant departments and lawyers.

[0135] Gathering feedback and updating the model

[0136] The server records the results of each reminder activity (e.g., email open rates and phone call response rates) and collects them as feedback data. The collected feedback data is used to retrain the analysis model and improve the accuracy of the next reminder plan.

[0137] Specific examples

[0138] For example, the server retrieves debtor A's payment history for the past year from the bank API and stores the data in JSON format in MongoDB. Next, it uses the Pandas library to fill in missing values ​​in the data and uses Scikit-learn to predict payment trends using a random forest model. Based on the analysis results, it determines the optimal method of reminding debtor A, automatically generates and sends a reminder email on the 15th. If there is no response, it makes a reminder phone call using Twilio's API. The results are recorded in the database and reflected in the next model update.

[0139] Prompt Sentence Examples

[0140] "Based on Debtor A's payment history, email communication history, and phone call history for the past year, please predict the risk of delay in the next debt payment and plan the optimal method and timing of reminder payments."

[0141] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0142] Program processing steps

[0143] Step 1: Data collection

[0144] Subject: Server

[0145] The server receives the debtor's identification information as input. The server collects data such as the debtor's payment history, communication history, and financial status through the APIs of credit bureaus, banks, and telecommunications companies. For example, the server obtains the payment history of debtor A through an API call and saves the data in JSON format. During this process, the server stores the obtained data in a database.

[0146] Input: Debtor Identification Information

[0147] Output: Data on debtor's payment history, communication history, financial status, etc. (JSON format)

[0148] Step 2: Data Preprocessing

[0149] Subject: Server

[0150] The server receives the collected data as input. It uses the Pandas library to preprocess the data, removing noise, imputing missing values, and preparing it in a format suitable for analysis. During this process, it generates a data frame and performs operations such as removing outliers.

[0151] Input: Collected data (JSON format)

[0152] Output: Preprocessed data (data frame)

[0153] Step 3: Data analysis

[0154] Subject: Server

[0155] The server receives preprocessed data as input. It uses Scikit-learn to apply machine learning algorithms (e.g., random forests) to predict payment trends and optimal reminder timing. The analysis results are output as predictions for each debtor, including the likelihood of the next late payment and the optimal reminder method.

[0156] Input: Preprocessed data (data frame)

[0157] Output: Analysis results (payment trends, optimal reminder timing)

[0158] Step 4: Generate a customized reminder plan

[0159] Subject: Server

[0160] The server receives the analysis results as input. Based on this, it generates a customized reminder plan for each debtor. For example, if the analysis shows that email reminders are effective for debtor A, the server determines the content and timing of the emails.

[0161] Input: Analysis results (payment trends, optimal reminder timing)

[0162] Output: Customized reminder plan

[0163] Step 5: Run automatic reminders

[0164] Subject: Server

[0165] The server takes your customized reminder plan as input and automatically executes reminder activities at the specified dates and times, for example, generating and sending reminder emails using an SMTP server, and optionally placing automated calls using Twilio's APIs to play pre-recorded messages.

[0166] Input: Customized reminder plan

[0167] Output: Reminder actions taken (emails sent, call logs)

[0168] Step 6: Track down the missing person

[0169] Subject: Server

[0170] The server receives contact information for unreachable debtors as input, and uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information and update the database.

[0171] Input: Unreachable person's contact information

[0172] Output: Latest updated contact information

[0173] Step 7: Legal assistance

[0174] Subject: Server

[0175] The server receives the necessary legal procedure information as input, and then automatically generates mediation and bankruptcy procedure documents based on templates and sends them to the relevant departments and lawyers.

[0176] Input: Legal process information and templates

[0177] Output: Auto-generated legal document

[0178] Step 8: Gather feedback and update the model

[0179] Subject: Server

[0180] The server receives as input the results of each reminder activity (e.g., email open rates and phone response rates), which are collected as feedback data and used to retrain the analytical model, improving the accuracy of the next reminder plan.

[0181] Input: Results of reminder activities (open rate, response status)

[0182] Output: Updated analytical model

[0183] These are the specific processing steps of this system's program. Through this process, collection activities can be optimized for each debtor, greatly improving the efficiency of debt collection.

[0184] (Application example 1)

[0185] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0186] In the debt collection process, debtors frequently experience payment delays and are unable to be contacted, resulting in a decline in collection efficiency. Other issues include the time and cost required to simultaneously contact multiple debtors and the preparation of legal paperwork. Furthermore, it can be difficult to determine the optimal timing for collection, which can result in the ineffectiveness of payment collection efforts. These issues need to be resolved.

[0187] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0188] In this invention, the server includes means for collecting data such as debtor payment history, communication history, and financial status, means for analyzing the collected data to generate an optimal collection plan for each debtor, means for automatically carrying out collection activities for debtors based on the generated collection plan, means for searching for and tracking the latest contact information for debtors who have become unreachable, means for generating documents to support necessary legal procedures, means for collecting the results of the collection activities and updating the model based on the analysis results, means for predicting payment trends and determining the optimal timing for collection activities to prevent payment delays, and means for automatically generating and sending notification messages at specified dates and times. This streamlines the debt collection process and enables fast and effective debt collection activities.

[0189] A "debtor" is an individual or corporation that is obligated to borrow money from a financial institution, business partner, etc., or to pay for goods or services.

[0190] A "payment history" is a record of past payments made by a debtor, including details such as the date, amount, and method of payment within a specific period of time.

[0191] "Communication history" refers to a record of the debtor's communication activities, such as phone calls, emails, and messages, including the date, time, frequency, and content of the messages sent and received.

[0192] "Financial situation" refers to information that shows the debtor's economic status, such as income, expenses, assets, and liabilities.

[0193] "Means of collecting data" refers to the mechanisms and technologies for collecting the necessary data using APIs from credit information agencies, bank transaction databases, telecommunications company APIs, etc.

[0194] "Means for analyzing data" refers to methods and techniques for preprocessing collected data and using machine learning models to predict payment trends and find the optimal timing for reminders.

[0195] A "demand collection plan" is a plan that specifically outlines when and how to request payment from each debtor.

[0196] "Means for carrying out collection activities" refers to systems or technologies that automatically notify or contact debtors based on the generated collection plan.

[0197] "Means for searching and tracking contact information" refers to methods and technologies for identifying the latest address and contact information of a debtor who has lost contact, using resident registration databases or telecommunications company APIs.

[0198] "Means for generating documents to support legal procedures" refers to systems and technologies that automatically create documents necessary for legal responses such as arbitration procedures and bankruptcy procedures.

[0199] "Means for collecting the results of prompting activities and updating the model" refers to methods and techniques for evaluating the effectiveness of prompting activities, improving the analytical model based on that data, and increasing the accuracy of future prompting plans.

[0200] "Measures to prevent payment delays" refers to methods and technologies that use machine learning models to predict payment trends and take appropriate action in advance if there is a high possibility of a delay.

[0201] "Means for generating and sending notification messages" refers to systems or technologies for automatically creating and sending payment reminder messages to debtors at specified dates and times.

[0202] The following describes an embodiment of the present invention.

[0203] The server first collects data such as the debtor's payment history, communication history, and financial situation. In this step, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the communication history is obtained from a telecommunications company's API.

[0204] The server then analyzes the collected data. During data analysis, the data is preprocessed to remove noise and fill in missing values. After that, a machine learning algorithm is used to predict payment trends and the optimal timing for reminders. Specifically, past delay patterns are input into the model to predict the likelihood of the next delay.

[0205] Based on the analysis results, the server generates a customized reminder plan for each debtor. This plan includes the optimal reminder method (SNS notification, app notification, email) and the optimal timing. For example, if past data shows that a particular debtor is likely to respond to SNS notifications, the server will create a plan to first remind that debtor via SNS notification.

[0206] Based on the generated reminder plan, the server automatically generates and sends notification messages at the specified date and time. For example, when a payment deadline approaches, a notification message is automatically sent to the debtor's smartphone. The server also records the notification opening rate and payment completion rate to evaluate the effectiveness of the notifications.

[0207] For debtors who have been lost contact, the server tracks their latest contact information, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact details, allowing for lost contact to be reestablished and effective collection efforts to continue.

[0208] In addition, the server automatically generates documents required for mediation procedures, bankruptcy procedures, etc. to assist with necessary legal procedures, and the generated documents are sent to relevant departments and lawyers.

[0209] The server collects the results of debt collection activities and updates the model to reflect them in the next collection plan. For example, it analyzes email open rates and phone response status, and improves the accuracy of the model through a feedback loop.

[0210] Hardware and software used

[0211] In this system, software is installed on the server to collect data via the credit information agency's API, the bank's transaction database, and the telecommunications company's API. Python libraries such as pandas and scikit-learn are used for data analysis, and smtplib is used to generate and send notification messages. RandomForestClassifier is used as the machine learning model.

[0212] Specific examples

[0213] For example, the payment history of a certain debtor A over the past year is collected, and a machine learning model is used to predict the likelihood of the next payment being late. Based on the analysis results, an SNS notification is selected for Debtor A, and a notification message is sent the day before the payment due date. This notification message is automatically generated, and the open rate and response status after sending are recorded. This data is reflected in the next reminder plan.

[0214] Prompt Sentence Examples

[0215] "Please predict the likelihood of User A's next payment being late based on their bank payment history, communication history, and financial situation over the past year. Based on the results, please provide a Python script that will generate an optimal reminder plan and automate the process of sending a notification message to User A."

[0216] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0217] Step 1:

[0218] Data collection: The server collects data such as the debtor's payment history, communication history, and financial situation via the API of credit information agencies, bank transaction databases, telecommunications company APIs, etc. Specifically, for a specific debtor, the payment history for the past year is obtained from the bank database, and the communication history is obtained from the telecommunications company API.

[0219] Input: Debtor identification information (e.g., Debtor ID)

[0220] Output: Datasets such as payment history, communication history, and financial status

[0221] Step 2:

[0222] Data preprocessing: The server preprocesses the collected data, removing noise and imputing missing values. This preprocessing cleans the data for later analysis. For example, missing values ​​are imputed with the mean value and outliers are removed.

[0223] Input: Collected dataset

[0224] Output: A preprocessed and clean dataset

[0225] Step 3:

[0226] Data analysis: The server analyzes the preprocessed data. It uses the machine learning algorithm RandomForestClassifier to predict payment trends and the optimal timing for reminders. It also inputs past delay patterns into the model to predict the likelihood of the next delay.

[0227] Input: Preprocessed clean dataset

[0228] Output: Predictions regarding payment trends and likelihood of delays

[0229] Step 4:

[0230] Generation of reminder plan: Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, if it is determined that a particular debtor is likely to respond to SNS notifications, it will create a reminder plan for that debtor via SNS notifications.

[0231] Input: Prediction result

[0232] Output: Optimal reminder plan

[0233] Step 5:

[0234] Generation and sending of notification messages: The server automatically generates and sends notification messages at the specified date and time based on the generated reminder plan. For example, a reminder message can be sent to the debtor's smartphone the day before the payment deadline.

[0235] Input: Optimal reminder schedule

[0236] Output: Notification message sent

[0237] Step 6:

[0238] Contact information tracking: For debtors who have become unable to contact us, the server uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, thereby restoring lost contact.

[0239] Input: Uncontactable debtor information

[0240] Output: Latest contact information

[0241] Step 7:

[0242] Generation of documents to support legal procedures: The server automatically generates documents required for necessary legal procedures (e.g., arbitration procedures, bankruptcy procedures) and sends them to the relevant departments and lawyers.

[0243] Input: Debtor's Legal Information

[0244] Output: Auto-generated legal document

[0245] Step 8:

[0246] Model update: The server collects the results of the reminder activities and updates the analytical model based on them. This improves the accuracy of the next reminder plan. For example, the server records the opening rate of notification messages and the payment completion rate and reflects them in the model.

[0247] Input: Result data of reminder activity

[0248] Output: Updated analytical model

[0249] Furthermore, an emotion engine that estimates the user's emotion may be combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59 and perform identification processing using the user's emotion.

[0250] 1. Data Collection

[0251] Subject: Server

[0252] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains the payment history for the past year from the bank database, and the communication history (phone and email records) from the telecommunications company's API.

[0253] 2. Data Analysis

[0254] Subject: Server

[0255] The server preprocesses the collected data, removing noise and filling in missing values. It then uses machine learning algorithms to predict payment trends and the optimal timing for reminders. The server inputs past delay patterns into the model and calculates the risk of the next delay.

[0256] 3. Generate a customized reminder plan

[0257] Subject: Server

[0258] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data indicates that a debtor is likely to respond to email reminders, the server will first plan to remind that debtor by email.

[0259] 4. Utilizing the Emotion Engine

[0260] Subject: Server

[0261] The server uses an emotion engine to recognize the user's (debtor's) emotions during collection activities. The emotion engine detects emotions through voice and text analysis and identifies emotional states such as positive, negative, and neutral. For example, during a phone call, the server analyzes emotions from the user's tone of voice and the content of their text replies.

[0262] 5. Execute automatic reminders

[0263] Subject: Server

[0264] The server automatically executes reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. It also makes reminder calls via an automated calling system and plays recorded messages as needed. The emotion information recognized by the emotion engine is used to adjust the reminder plan.

[0265] 6. Tracking Lost Contacts

[0266] Subject: Server

[0267] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest address and contact information using a resident registration database or a telecommunications company's API. For example, it accesses a resident registration database to obtain the debtor's latest address information.

[0268] 7. Legal assistance

[0269] Subject: Server

[0270] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for sending to the relevant departments and lawyers.

[0271] 8. Feedback and optimization

[0272] Subject: Server

[0273] The server collects the results of collection activities and records them in a database. Information such as the open rate of emails sent, response status of phone calls, and whether or not payments were made is saved as a log. In addition, emotional information recognized by the emotion engine is also collected, and the generative AI model is updated based on the analysis results. This feedback loop improves the accuracy of collection plans for future collections, enabling more efficient debt collection.

[0274] Examples:

[0275] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0276] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and allows for more efficient and faster debt collection activities while taking into account the emotional state of the user.

[0277] The processing flow will be explained below.

[0278] Step 1:

[0279] Subject: Server

[0280] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains Debtor A's payment history for the past year from the bank database, and collects Debtor A's phone and email history for the past six months through the telecommunications company's API.

[0281] Step 2:

[0282] Subject: Server

[0283] The server preprocesses the collected data, which includes cleaning the data (removing noise) and imputing missing values. For example, the server removes incomplete data rows, imputes missing values ​​with the mean or median, and converts non-standardized data into a standardized format.

[0284] Step 3:

[0285] Subject: Server

[0286] The server analyzes the preprocessed data and extracts payment trends for each borrower. This analysis uses machine learning models to detect patterns of late payments from borrowers based on their past payment history. For example, the server can identify borrowers who have frequently made late payments in the past and predict the risk of the next late payment.

[0287] Step 4:

[0288] Subject: Server

[0289] The server generates an optimal reminder plan for each debtor. The generated plan includes the optimal reminder method (telephone, email, written) and optimal timing. Specifically, if a debtor has responded well to email reminders in the past, the server generates a plan that prioritizes email reminders.

[0290] Step 5:

[0291] Subject: Server

[0292] The server uses an emotion engine to recognize the user's emotions during the prompting activity. The emotion engine detects emotions through voice analysis and text analysis. For example, during a phone prompting call, the server can determine the user's emotional state (positive, negative, neutral, etc.) from the tone and speed of the user's voice and the content of the message.

[0293] Step 6:

[0294] Subject: Server

[0295] The server adjusts the reminder plan in real time based on the user's emotional information recognized by the emotion engine. For example, if the user shows negative emotions, the server may soften the tone of the reminder or temporarily suspend the reminder.

[0296] Step 7:

[0297] Subject: Server

[0298] The server automatically carries out reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder phone calls via an automated calling system and plays recorded messages. For example, the server automatically generates and sends a reminder email to debtor A on the 15th, and if there is no response, it makes a reminder phone call.

[0299] Step 8:

[0300] Subject: Server

[0301] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest addresses and contact information using a resident registration database or a telecommunications company's API. For example, it accesses the resident registration database to obtain the latest address information of Debtor A.

[0302] Step 9:

[0303] Subject: Server

[0304] The server automatically generates documents to support the necessary legal procedures. Mediation procedure documents and bankruptcy procedure documents are automatically created based on templates and sent to the relevant departments and lawyers. For example, the server automatically generates mediation procedure documents for Debtor A and prepares them to be sent to the lawyer.

[0305] Step 10:

[0306] Subject: Server

[0307] The server collects the results of the reminder activities and records them in a database. It stores information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made as logs. It also collects emotional information recognized by the emotion engine.

[0308] Step 11:

[0309] Subject: Server

[0310] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and apply new patterns and trends to it.

[0311] Example 2

[0312] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0313] Conventional debt collection systems have the problem of being inefficient and time-consuming, requiring a lot of manual work. It is also difficult to implement collection activities that take into account the debtor's feelings, and there is a lack of appropriate timing and means for approaching debtors. Furthermore, tracking down debtors who have become unreachable and generating the necessary documents for legal procedures are often done manually, which is labor-intensive and time-consuming. There is a need for a system that can solve these problems and achieve efficient and effective debt collection.

[0314] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0315] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial situation; means for preprocessing the collected data to remove noise and fill in missing values; means for analyzing the preprocessed data and generating an optimal collection plan for each debtor; means for recognizing the debtor's emotions based on the analysis results and generating a customized collection plan; means for automatically conducting loan collection activities based on the generated collection plan; means for tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of the collection activities and updating the generation AI model based on the analysis results. This automates the debt collection process and enables efficient and prompt debt collection activities that take into account the user's emotional state.

[0316] "Data collection tools" refers to hardware and software used to obtain data such as debtors' payment history, communication history, and financial status.

[0317] The "preprocessing means" refers to a software module that has the function of removing noise and filling in missing values ​​from collected data.

[0318] "Data Analysis Tools" means machine learning algorithms and related software used to analyze pre-processed data and predict debtor behavior patterns and optimal collection strategies.

[0319] The "demand plan generation means" refers to software and algorithms that determine the means and timing of demands customized for each debtor based on the analysis results and generate a plan.

[0320] "Emotion Recognition Means" means software modules and algorithms for recognizing the emotional state of a debtor through speech and text analysis.

[0321] "Automated collection measures" refers to software and hardware that automates loan collection activities using email and automated calling systems based on generated collection plans.

[0322] "Tracking means" refers to software and hardware that has the ability to obtain the latest contact information of debtors who have become unreachable by using resident registration databases or telecommunications company APIs.

[0323] "Document Generation Measures" means software modules and templates for automatically generating arbitration documents and bankruptcy documents to support necessary legal proceedings.

[0324] The "feedback collection means" is a software module that collects the results of prompting activities and records them as a log in a database.

[0325] "Generative AI model update means" refers to a means for updating the generative AI model based on the results of analysis and prompting activities, thereby improving the accuracy of future prompting plans.

[0326] This invention is a system for streamlining debt management and collection operations, collecting necessary information from multiple data sources, automating analysis and collection activities, and further performing emotional analysis, enabling flexible responses according to the emotional state of the debtor.

[0327] Data collection

[0328] The server collects data such as the debtor's payment history, communication history, and financial situation. Specifically, it obtains the necessary information using the credit bureau's API, the bank's transaction database, and the telecommunications company's API. The server sends requests to these data sources, for example, accessing the "credit bureau API" endpoint to obtain the debtor's payment history. Similarly, it uses an SQL query to obtain transaction history for the past year from the "bank transaction database," and obtains phone and email records via the telecommunications company's API. This data is then stored in a "data collection folder" by the server.

[0329] Data analysis and preprocessing

[0330] The server preprocesses the collected data, removing noise and filling in missing values. It uses a "data cleaning function" to fill in missing values ​​and applies a "noise removal algorithm" to refine the data. After this process is complete, the preprocessed data is input into a "machine learning model" to predict payment trends and delay risk. Specifically, it uses a "payment pattern identification model" to analyze past payment data and calculate the risk of the next delay.

[0331] Generate a customized reminder plan

[0332] Based on the analysis results, the server generates the optimal reminder plan for each debtor. To generate the plan, a "plan generation algorithm" is used to determine the optimal reminder method (telephone, email, written) and timing. For example, in the case of "Debtor A," if past data shows that the debtor has a high response rate to email reminders, email is selected as the top priority method and entered into the reminder plan table.

[0333] Utilizing the Emotion Engine

[0334] The server uses an "emotion engine" to recognize the debtor's emotions during collection activities. This engine detects emotions by applying a "voice analysis module" to phone recordings and a "text analysis module" to email and chat content. Emotional states (positive, negative, neutral) are recorded in an "emotion log" to help adjust collection plans.

[0335] Execute automatic reminders

[0336] The server automatically executes reminder activities based on the generated reminder plan. The "reminder email generation module" generates the email content and sends it at the specified date and time using the "email sending script." It also makes reminder calls via the "automatic call system" and plays back recorded messages. The reminder plan is adjusted in real time using emotional information recognized by the emotion engine.

[0337] Tracking lost contacts

[0338] The server tracks the latest contact information of unreachable debtors using the resident registration database and the API of the telecommunications company. Specifically, it sends a request to the "resident registration API" to obtain the latest address and contact information. This information is then updated in the "contact database."

[0339] Legal process assistance

[0340] The server automatically generates documents to support the necessary legal procedures. Mediation and bankruptcy documents are created based on the "document generation template" and formatted by the "PDF generation module." The documents are then ready to be sent to the relevant departments and lawyers.

[0341] Feedback and Optimization

[0342] The server collects the results of the reminder activities and records them in a "feedback database." For example, it stores logs of the open rate of emails sent, response status of phone calls, and whether or not payments were made. It also collects emotional information recognized by the emotion engine and updates the "generative AI model." This feedback loop improves the accuracy of future reminder plans.

[0343] Examples:

[0344] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0345] Example prompt sentence:

[0346] "Calculate the optimal method and timing for reminding Debtor A based on their payment history and communication history over the past year. Then, optimize it using an emotion engine."

[0347] The above is a specific embodiment for carrying out the invention. This system automates the debt collection process and enables efficient and fast debt collection activities that take into account the emotional state of the user.

[0348] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0349] Step 1:

[0350] The server collects data such as the debtor's payment history, communication history, and financial situation. The API of the credit information agency, the bank's transaction database, and the API endpoint of the telecommunications company are used as input. The server then sends requests to each API, obtains the required data, and saves it in a data collection folder. Specifically, it sends an API request, converts the returned data into a specified format, and saves it.

[0351] Step 2:

[0352] The server preprocesses the collected data. The input is the various collected data, and based on that, it performs noise removal and missing value imputation. It uses data cleaning functions to properly impute missing values ​​and applies noise removal algorithms. The output of this step is preprocessed data, which is sent to the next analysis step. Specific operations include filtering outliers and imputing missing values ​​using statistical methods.

[0353] Step 3:

[0354] The server analyzes the preprocessed data. The input is the preprocessed data, and the payment trends and late payment risk are analyzed based on a machine learning model. The received data is analyzed using a payment pattern identification model to calculate the late payment risk for each debtor. The results are output as an analysis report and used in the next step. Specifically, past payment data is input into the model to calculate the predicted late payment risk.

[0355] Step 4:

[0356] The server generates a customized reminder plan based on the analysis results. The input is the analysis report, and the plan generation algorithm is used to determine the individual reminder methods and timing. The generated plan is saved in the reminder plan table and sent to the execution step. Specifically, the server sets the optimal approach method and timing for each debtor based on the analysis results obtained from the database.

[0357] Step 5:

[0358] The server automatically executes reminder activities based on the generated reminder schedule. The input is data from the reminder schedule table, and the reminder email generation module generates the email content and sends it at the date and time specified by the email sending script. It also executes telephone reminders via an automatic calling system and plays back recorded messages. Specifically, it automatically executes the email script and sends an emergency message via the telephone system.

[0359] Step 6:

[0360] The server tracks the latest contact information for unreachable debtors. The input is the debtor information to be tracked, and the latest address and contact information is obtained using the resident registration database and the API of the telecommunications company. The server then updates the contact information in the contact database with the new information. Specifically, it sends a request to the resident registration API and updates the database with the obtained information.

[0361] Step 7:

[0362] The server automatically generates documents to support the necessary legal procedures. The input is the data required for the legal procedures, and mediation procedure documents and bankruptcy procedure documents are created using document generation templates and formatted in the PDF generation module. The generated documents are sent to relevant departments and lawyers. Specifically, the required information is entered into the document template, which is then automatically generated in PDF format.

[0363] Step 8:

[0364] The server collects the results of the prompting activities and updates the model. The inputs are the results of the prompting activities (email open rate, phone response status, etc.) and emotion recognition results, and the logs are saved in a feedback database. Based on this, the generative AI model is updated to improve the accuracy of future prompting plans. Specifically, the analysis results are added to the model learning data, and the model is retrained to improve accuracy.

[0365] The above are the specific processing steps of this system.

[0366] (Application example 2)

[0367] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart device 14 will be referred to as a "terminal."

[0368] In today's world, effective debt collection activities through communication with debtors are important, but traditional methods fail to fully consider the individual circumstances and emotions of debtors. Furthermore, the advertising industry faces challenges in accurately capturing users' interests and purchasing intent and delivering advertisements at the optimal time. To solve these challenges, a system is needed that analyzes the behavior and emotions of debtors and users and automatically responds optimally.

[0369] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0370] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for analyzing the collected data to generate an optimal collection schedule for each debtor; means for automatically carrying out collection activities for debtors based on the generated collection schedule; means for searching for and tracking the latest contact information for debtors who have become unreachable; means for generating documents to support necessary legal procedures; means for collecting the results of the collection activities and updating a model based on the analysis results; means for collecting user purchase history, browsing history, and communication history; means for preprocessing the collected data and analyzing user interests and purchasing intentions using a machine learning algorithm; means for generating an optimal advertising plan for each user based on the analysis results; means for analyzing user emotions using an emotion analysis engine and adjusting advertisements; means for automatically delivering advertisements to users based on the generated advertising plan; and means for collecting the results of advertisement delivery and optimizing subsequent advertising plans. This automates the debt collection process and advertisement delivery process, enabling more efficient and effective responses while taking into account the user's emotional state and behavioral patterns.

[0371] A "debtor" is a person who has borrowed money from a financial institution or money lender and is obligated to repay the loan.

[0372] "Payment history" refers to a record of the payments a debtor has made to date, including the date, time, amount, and method of each payment.

[0373] "Communication history" refers to a record of contacts and conversations made through the means of communication used by the debtor (such as telephone or email).

[0374] "Financial situation" refers to the debtor's financial condition, including income, expenses, assets, and liabilities.

[0375] "Data collection methods" refer to the methods and techniques used to compile the required information, including APIs and databases.

[0376] "Data analysis tools" refer to techniques and methods used to pre-process and subsequently analyze collected data, including, for example, noise reduction and machine learning algorithms.

[0377] A "repayment plan" refers to specific methods and schedules for encouraging debtors to repay.

[0378] "Automatic collection activity means" refers to a technique or method for automatically sending collections to debtors based on the generated collection plan.

[0379] "Contact information retrieval means" refers to a technique or method for identifying the most current contact information for a debtor who has become unreachable.

[0380] "Document generation means for supporting legal procedures" refers to techniques and methods for automatically creating documents to be used in necessary legal procedures.

[0381] "Model update methods" refer to techniques and methods for updating generative algorithms and machine learning models based on analysis results, improving accuracy from the next time onwards.

[0382] "Purchase history" refers to a list or record of products a user has purchased in the past.

[0383] "Browsing history" refers to a record of the web pages and content a user has viewed on the Internet.

[0384] A "machine learning algorithm" is a mathematical model or computational procedure used to make predictions or classifications based on collected data.

[0385] An "advertising plan" refers to a specific method and schedule for delivering optimal advertisements to users.

[0386] An "emotion analysis engine" is software or algorithms for detecting and analyzing a user's emotional state.

[0387] "Automatic advertisement distribution means" refers to a technique or method for automatically distributing advertisements to users in accordance with a generated advertising plan.

[0388] "Advertising delivery result optimization means" refers to techniques and methods for optimizing future advertising plans based on the results of advertising delivery.

[0389] System Overview

[0390] The system of the present invention automates the collection and analysis of data such as debtor payment history, communication history, and financial situation, and has the function of generating an optimal collection plan and automatically carrying out collection activities.It can also be applied to the advertising field, and has the function of delivering optimal advertisements based on users' purchase history, browsing history, and emotional information.The system consists of the following elements: server, terminal, and user.

[0391] Program Details

[0392] Hardware and software used

[0393] Hardware: Servers, smartphones, smart glasses, head-mounted displays

[0394] Software: Emotion engine (Emotion API), machine learning model (TensorFlow), database (MySQL), ad delivery management software (Ad Manager API)

[0395] Data collection

[0396] The server uses APIs and databases to collect data on the debtor's payment history, communication history, and financial status. For example, it obtains payment history for the past year from a bank database and call history from a telecommunications company's API.

[0397] Data analysis

[0398] The server preprocesses the collected data, removing noise and filling in missing values, and then analyzes the data using a machine learning algorithm (using TensorFlow) to predict the debtor's payment trends and the optimal timing for collection.

[0399] Generate reminder schedule

[0400] Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, for debtors for whom email reminders are effective, email reminders are sent first, and if there is no response, the server switches to phone reminders.

[0401] Utilizing the Emotion Engine

[0402] The server uses the emotion engine to analyze the debtor's emotions during collection activities, for example, by analyzing the debtor's tone of voice during a collection call, and if negative emotions are detected, adjust the collection plan accordingly.

[0403] Execute automatic reminders

[0404] Based on the generated reminder plan, the server automatically executes reminder activities, automatically generates reminder emails and sends them at the specified date and time, and also makes reminder calls via an automated calling system and plays recorded messages.

[0405] Collection of user purchasing and browsing history

[0406] The server collects the user's purchase history and browsing history using various APIs, such as the History API of a web browser or the API of a communication app.

[0407] Advertising plan generation and ad delivery

[0408] The server preprocesses the collected user data and uses machine learning algorithms to analyze the user's interests and purchasing intent. Based on the analysis results, it generates an optimal advertising plan for each user. Based on the generated advertising plan, the server automatically delivers advertisements to the user.

[0409] Result feedback and model optimization

[0410] The server records the results of ad delivery in a database and optimizes future ad plans. Ad click rates and purchase rates are fed back into the model to improve the ad delivery algorithm.

[0411] Specific examples

[0412] The server retrieves debtor A's payment history for the past year from the bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th of each month, and if there is no response, a reminder phone call is made. Similarly, the server analyzes user B's past purchase and browsing history and delivers the optimal advertisement during the lunch break.

[0413] Prompt Sentence Examples

[0414] "We analyze a user's purchasing and browsing history over the past year to predict what products they are likely to purchase next. We use the Emotion API to analyze their emotions while watching an ad, and then display ads from categories that they responded positively to."

[0415] This will automate the debt collection and advertising delivery processes, enabling more efficient and effective operations.

[0416] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0417] Step 1:

[0418] Data collection

[0419] The server uses APIs and databases to collect the necessary data.

[0420] Input: Bank database API, Telecommunications company API, Web browser History API, Telecommunications app API

[0421] Specific operation: Obtain the debtor's payment history for the past year from the bank database, obtain call history from the telecommunications company's API, and obtain the user's purchase history and browsing history using the web browser's History API and the communication app's API.

[0422] Output: Data set of payment history, communication history, purchase history, browsing history, etc. for the relevant debtors and users

[0423] Step 2:

[0424] Data Preprocessing

[0425] The server preprocesses the collected data, removing noise and completing missing values.

[0426] Input: The dataset collected in step 1

[0427] Specific actions: Identify and remove or correct inaccurate information in collected data, and impute missing values ​​in an appropriate manner.

[0428] Output: A clean and reliable dataset

[0429] Step 3:

[0430] Data analysis

[0431] The server analyzes the pre-processed data using machine learning algorithms to predict the debtor's payment trends and the optimal timing for reminders.

[0432] Input: The dataset preprocessed in step 2, the machine learning algorithm (TensorFlow)

[0433] How it works: It uses machine learning algorithms to analyze each debtor's payment habits and risk of late payments, and also predicts the next likely purchase based on the user's purchase and browsing history.

[0434] Output: Analysis results on payment trends, optimal reminder timing, user interests, and purchasing intent

[0435] Step 4:

[0436] Generate promotional and advertising plans

[0437] Based on the analysis results, the server generates an optimal reminder plan for each debtor and an optimal advertising plan for each user.

[0438] Input: Analysis results obtained in Step 3

[0439] Specific operation: Based on the analysis results, the system determines the optimal reminder method (telephone, email, written) and timing for each debtor and generates a reminder plan. It also determines the optimal advertising content and distribution timing for each user and generates an advertising plan.

[0440] Output: Reminder plan for each debtor, advertising plan for each user

[0441] Step 5:

[0442] Emotion analysis

[0443] The server uses an emotion engine to analyze the user's emotions during the promotional activities and advertisement viewing.

[0444] Input: Voice data during promotional activities and ad viewing, text data, emotion engine (Emotion API)

[0445] Specific operation: The emotion engine analyzes voice and text data to identify emotional states such as positive, negative, or neutral.

[0446] Output: Analysis results on user sentiment

[0447] Step 6:

[0448] Execution of automatic reminders and advertisement distribution

[0449] The server automatically executes promotional activities and advertisement distribution based on the generated plan.

[0450] Input: Promotion plan and advertising plan generated in step 4

[0451] Specific operations: Automatically generate and send reminder emails to debtors at the specified date and time. Also, make reminder phone calls via an automated calling system. Furthermore, distribute advertisements at the specified times based on the advertising plan.

[0452] Output: Reminders performed and ads delivered

[0453] Step 7:

[0454] Results feedback and optimization

[0455] The server collects the results of the promotional activities and advertisement distribution and optimizes the plan for the next time.

[0456] Input: Promotional activity result data, ad delivery result data

[0457] Specific operations: Results such as the open rate of reminder emails, response status of phone calls, click rates and purchase rates of ads are recorded in a database, and the model is retrained and optimized.

[0458] Output: Updated machine learning models and optimization data for next promotional and advertising plans

[0459] Through these steps, the system of the invention automates the debt collection and advertising distribution processes, achieving effective and efficient operations.

[0460] The specific processing unit 290 transmits the result of the specific processing to the smart device 14. In the smart device 14, the control unit 46A causes the output device 40 to output the result of the specific processing. The microphone 38B acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 38B to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0461] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0462] In the above embodiment, an example in which the specific process is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific process may be performed by the smart device 14.

[0463] [Second embodiment]

[0464] FIG. 3 shows an example of the configuration of a data processing system 210 according to the second embodiment.

[0465] 3, the data processing system 210 includes the data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0466] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0467] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication I / F 44. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, and the camera 42 are also connected to the bus 52.

[0468] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0469] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0470] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0471] Fig. 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Fig. 4, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0472] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0473] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0474] In the smart glasses 214, the reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0475] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the smart glasses 214 will be referred to as the "terminal."

[0476] 1. Data Collection

[0477] Subject: Server

[0478] The server first collects data such as the debtor's payment history, communication history, and financial situation. At this stage, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the email communication history is obtained from the email server.

[0479] 2. Data Analysis

[0480] Subject: Server

[0481] The server analyzes the collected data. When analyzing the data, it first preprocesses it to remove noise and fill in missing values. It then uses a machine learning algorithm to predict payment trends and the optimal timing for reminders. Specifically, it inputs past delay patterns into a model to predict the likelihood of the next delay.

[0482] 3. Generate a customized reminder plan

[0483] Subject: Server

[0484] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will create a plan to first remind that debtor by email.

[0485] 4. Execute automatic reminders

[0486] Subject: Server

[0487] The server automatically executes reminder activities based on the generated reminder plan. For example, it automatically generates and sends reminder emails at a specified date and time. If a reminder call is required, it makes a call via an automated calling system and plays a recorded message.

[0488] 5. Tracking Lost Contacts

[0489] Subject: Server

[0490] The server tracks the latest contact information of debtors who have become unreachable, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, allowing lost contact to be restored.

[0491] 6. Legal assistance

[0492] Subject: Server

[0493] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[0494] 7. Feedback and optimization

[0495] Subject: Server

[0496] The server collects the results of collection activities and updates the model. For example, it records the open rate of emails and the response status of phone calls, and reflects this in the next collection plan. This feedback loop improves the accuracy of future collection plans, enabling more efficient debt collection.

[0497] Examples:

[0498] For example, the server retrieves debtor A's payment history for the past year via the bank database and API, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a further reminder phone call is made. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0499] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and enables more efficient and faster debt collection activities.

[0500] The processing flow will be explained below.

[0501] Step 1:

[0502] Subject: Server

[0503] The server collects data such as the debtor's payment history, communication history, and financial situation. This collection process obtains the necessary information using the API of a credit information agency, a bank's transaction database, and a telecommunications company's API. For example, the server sends a query to a bank database to obtain a specific debtor's payment history for the past year. It also obtains communication history (phone and email records) from a telecommunications company's API.

[0504] Step 2:

[0505] Subject: Server

[0506] The server preprocesses the collected data, including cleaning the data (removing noise) and imputing missing values. Specific operations include deleting incomplete data rows and imputing missing values ​​with the mean or median. It also converts data in non-standard formats into a standard format.

[0507] Step 3:

[0508] Subject: Server

[0509] The server analyzes the preprocessed data and extracts payment trends for each debtor. This analysis uses machine learning models to detect debtor delay patterns from past payment history. For example, the server can identify debtors who have had many late payments in the past and calculate the risk of their next payment delay.

[0510] Step 4:

[0511] Subject: Server

[0512] The server generates an optimal reminder plan for each debtor. The generated reminder plan includes the optimal reminder method (phone, email, written) and optimal timing. For example, if a debtor has responded well to emails in the past, the server generates a plan that prioritizes email reminders.

[0513] Step 5:

[0514] Subject: Server

[0515] The server automatically executes reminder activities based on the generated reminder plan. Specifically, it automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder calls via an automated calling system and plays recorded messages.

[0516] Step 6:

[0517] Subject: Server

[0518] The server tracks the latest contact information of debtors who have become unreachable. To do this, it uses a resident registration database or a telecommunications company's API to identify the latest address and contact information. For example, the server accesses a resident registration database to obtain the debtor's latest address information.

[0519] Step 7:

[0520] Subject: Server

[0521] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for delivery to relevant departments and lawyers.

[0522] Step 8:

[0523] Subject: Server

[0524] The server collects the results of the collection activities and records them in a database, for example, by saving information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made.

[0525] Step 9:

[0526] Subject: Server

[0527] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and incorporate new patterns and trends.

[0528] Example 1

[0529] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0530] In conventional debt collection systems, collection plans for each debtor are general and uniform, and collections are not optimally tailored to the circumstances of each debtor. This reduces collection efficiency and makes it difficult to track debtors when they cannot be contacted. Furthermore, preparing documents required for legal procedures is a manual process that consumes time and resources.

[0531] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0532] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for preprocessing the collected data, removing noise, and completing missing values; means for analyzing the data using a machine learning algorithm and predicting the optimal timing and means of collection for each debtor; means for generating a collection plan customized for each debtor based on the prediction; means for automatically carrying out collection activities for debtors based on the generated collection plan; means for searching for and tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of collection activities and updating the model based on the analysis results. This automates the debt collection process, enables optimal collection activities tailored to the circumstances of individual debtors, efficient tracking when debtors cannot be contacted, and enables the rapid and accurate preparation of legal procedure documents.

[0533] A "debtor" is an individual or legal entity that has a financial obligation in the form of a debt, loan, or other form of debt.

[0534] "Payment history" is a record of past payments made by a debtor, including details such as due dates, amounts, and payment methods.

[0535] "Communication history" refers to a record of communication activities such as phone calls, emails, and messages between you and the debtor.

[0536] "Financial status" refers to information that indicates the debtor's current financial condition, and includes data on income, expenses, assets, liabilities, and the like.

[0537] "Data collection means" refers to a method or device by which the server obtains information such as the debtor's payment history, communication history, and financial situation.

[0538] "Data preprocessing" is the process of preparing collected data before analysis, such as removing noise from the data and filling in missing values.

[0539] "Machine learning algorithms" are statistical methods and models for finding patterns and knowledge in data.

[0540] An "analysis means" is a method or device for predicting a debtor's behavioral tendencies and the optimal timing for making a collection request by utilizing collected and preprocessed data.

[0541] A "demand plan" is a plan that determines the optimal means and timing of demands to prevent a debtor from delaying repayments.

[0542] A "customized reminder plan" is a reminder plan that is individually tailored based on each debtor's characteristics and behavior.

[0543] The "automatic prompting activity execution means" is a method or device that allows the server to automatically execute prompting means such as telephone calls or e-mails based on the generated prompting plan.

[0544] A "contact information tracking means" is a method or device for searching for and tracking the latest contact information of a debtor who has become unreachable.

[0545] "Legal procedure support means" refers to a method or device for generating documents related to necessary legal procedures and providing them to relevant departments and experts.

[0546] The "feedback collection means" is a method or device for collecting the results of prompting activities and reflecting them in the next prompting plan.

[0547] A "model updater" is a method or device for retraining an analytical model based on collected feedback data to improve its accuracy.

[0548] The present invention provides a system for automating and efficiently carrying out debt collection activities against specific debtors. A specific embodiment of this system will now be described.

[0549] Hardware and software used

[0550] This system operates in a server-based network environment. The specific hardware and software that supports each process in this system are listed below.

[0551] Server: Collects data, performs analytics, performs remediation activities, tracks contact information, assists with legal proceedings, gathers feedback, and updates the model.

[0552] Example of hardware used: A server equipped with a high-performance CPU and large-capacity RAM

[0553] Examples of software used: MongoDB (database), Pandas (data processing), Scikit-learn (machine learning), Twilio (automated calls), SMTP server (email sending)

[0554] Terminal: A device that interacts with a server and displays or inputs information.

[0555] Examples of hardware used: personal computers, tablets, smartphones

[0556] Specific processing of the system

[0557] Data collection

[0558] The server uses APIs from credit information agencies, banks, and telecommunications companies to collect data such as debtor payment history, communication history, and financial status. For example, the server obtains debtor A's payment history for the past year from the bank API and stores that data in MongoDB.

[0559] Data Preprocessing

[0560] The collected data is first preprocessed. The server uses the Pandas library to remove noise and impute missing values, then formats the collected data into a data frame.

[0561] Data analysis

[0562] Machine learning algorithms are applied to the preprocessed data. The server uses Scikit-learn and models such as random forests to predict the debtor's payment trends and the optimal timing for reminders. For example, it analyzes the debtor's delay patterns and predicts the likelihood of the next delay.

[0563] Generate a customized reminder plan

[0564] Based on the analysis results, the server generates a customized reminder plan for each debtor. This reminder plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will plan to first remind that debtor by email.

[0565] Execute automatic reminders

[0566] Based on the generated reminder plan, the server automatically executes reminder activities. It automatically generates reminder emails at the specified date and time and sends them via the SMTP server. In addition, if a phone reminder is required, it makes an automated call via Twilio's API and plays a pre-recorded message.

[0567] Tracking lost contacts

[0568] For debtors who can no longer be contacted, the server uses the resident registration database or the telecommunications company's API to search for the latest contact information and update the database.

[0569] Legal process assistance

[0570] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[0571] Gathering feedback and updating the model

[0572] The server records the results of each reminder activity (e.g., email open rates and phone call response rates) and collects them as feedback data. The collected feedback data is used to retrain the analysis model and improve the accuracy of the next reminder plan.

[0573] Specific examples

[0574] For example, the server retrieves debtor A's payment history for the past year from the bank API and stores the data in JSON format in MongoDB. Next, it uses the Pandas library to fill in missing values ​​in the data and uses Scikit-learn to predict payment trends using a random forest model. Based on the analysis results, it determines the optimal method of reminding debtor A, automatically generates and sends a reminder email on the 15th. If there is no response, it makes a reminder phone call using Twilio's API. The results are recorded in the database and reflected in the next model update.

[0575] Prompt Sentence Examples

[0576] "Based on Debtor A's payment history, email communication history, and phone call history for the past year, please predict the risk of delay in the next debt payment and plan the optimal method and timing of reminder payments."

[0577] The flow of the identification process in the first embodiment will be described with reference to FIG.

[0578] Program processing steps

[0579] Step 1: Data collection

[0580] Subject: Server

[0581] The server receives the debtor's identification information as input. The server collects data such as the debtor's payment history, communication history, and financial status through the APIs of credit bureaus, banks, and telecommunications companies. For example, the server obtains the payment history of debtor A through an API call and saves the data in JSON format. During this process, the server stores the obtained data in a database.

[0582] Input: Debtor Identification Information

[0583] Output: Data on debtor's payment history, communication history, financial status, etc. (JSON format)

[0584] Step 2: Data Preprocessing

[0585] Subject: Server

[0586] The server receives the collected data as input. It uses the Pandas library to preprocess the data, removing noise, imputing missing values, and preparing it in a format suitable for analysis. During this process, it generates a data frame and performs operations such as removing outliers.

[0587] Input: Collected data (JSON format)

[0588] Output: Preprocessed data (data frame)

[0589] Step 3: Data analysis

[0590] Subject: Server

[0591] The server receives preprocessed data as input. It uses Scikit-learn to apply machine learning algorithms (e.g., random forests) to predict payment trends and optimal reminder timing. The analysis results are output as predictions for each debtor, including the likelihood of the next late payment and the optimal reminder method.

[0592] Input: Preprocessed data (data frame)

[0593] Output: Analysis results (payment trends, optimal reminder timing)

[0594] Step 4: Generate a customized reminder plan

[0595] Subject: Server

[0596] The server receives the analysis results as input. Based on this, it generates a customized reminder plan for each debtor. For example, if the analysis shows that email reminders are effective for debtor A, the server determines the content and timing of the emails.

[0597] Input: Analysis results (payment trends, optimal reminder timing)

[0598] Output: Customized reminder plan

[0599] Step 5: Run automatic reminders

[0600] Subject: Server

[0601] The server takes your customized reminder plan as input and automatically executes reminder activities at the specified dates and times, for example, generating and sending reminder emails using an SMTP server, and optionally placing automated calls using Twilio's APIs to play pre-recorded messages.

[0602] Input: Customized reminder plan

[0603] Output: Reminder actions taken (emails sent, call logs)

[0604] Step 6: Track down the missing person

[0605] Subject: Server

[0606] The server receives contact information for unreachable debtors as input, and uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information and update the database.

[0607] Input: Unreachable person's contact information

[0608] Output: Latest updated contact information

[0609] Step 7: Legal assistance

[0610] Subject: Server

[0611] The server receives the necessary legal procedure information as input, and then automatically generates mediation and bankruptcy procedure documents based on templates and sends them to the relevant departments and lawyers.

[0612] Input: Legal process information and templates

[0613] Output: Auto-generated legal document

[0614] Step 8: Gather feedback and update the model

[0615] Subject: Server

[0616] The server receives as input the results of each reminder activity (e.g., email open rates and phone response rates), which are collected as feedback data and used to retrain the analytical model, improving the accuracy of the next reminder plan.

[0617] Input: Results of reminder activities (open rate, response status)

[0618] Output: Updated analytical model

[0619] These are the specific processing steps of this system's program. Through this process, collection activities can be optimized for each debtor, greatly improving the efficiency of debt collection.

[0620] (Application example 1)

[0621] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0622] In the debt collection process, debtors frequently experience payment delays and are unable to be contacted, resulting in a decline in collection efficiency. Other issues include the time and cost required to simultaneously contact multiple debtors and the preparation of legal paperwork. Furthermore, it can be difficult to determine the optimal timing for collection, which can result in the ineffectiveness of payment collection efforts. These issues need to be resolved.

[0623] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[0624] In this invention, the server includes means for collecting data such as debtor payment history, communication history, and financial status, means for analyzing the collected data to generate an optimal collection plan for each debtor, means for automatically carrying out collection activities for debtors based on the generated collection plan, means for searching for and tracking the latest contact information for debtors who have become unreachable, means for generating documents to support necessary legal procedures, means for collecting the results of the collection activities and updating the model based on the analysis results, means for predicting payment trends and determining the optimal timing for collection activities to prevent payment delays, and means for automatically generating and sending notification messages at specified dates and times. This streamlines the debt collection process and enables fast and effective debt collection activities.

[0625] A "debtor" is an individual or corporation that is obligated to borrow money from a financial institution, business partner, etc., or to pay for goods or services.

[0626] A "payment history" is a record of past payments made by a debtor, including details such as the date, amount, and method of payment within a specific period of time.

[0627] "Communication history" refers to a record of the debtor's communication activities, such as phone calls, emails, and messages, including the date, time, frequency, and content of the messages sent and received.

[0628] "Financial situation" refers to information that shows the debtor's economic status, such as income, expenses, assets, and liabilities.

[0629] "Means of collecting data" refers to the mechanisms and technologies for collecting the necessary data using APIs from credit information agencies, bank transaction databases, telecommunications company APIs, etc.

[0630] "Means for analyzing data" refers to methods and techniques for preprocessing collected data and using machine learning models to predict payment trends and find the optimal timing for reminders.

[0631] A "demand collection plan" is a plan that specifically outlines when and how to request payment from each debtor.

[0632] "Means for carrying out collection activities" refers to systems or technologies that automatically notify or contact debtors based on the generated collection plan.

[0633] "Means for searching and tracking contact information" refers to methods and technologies for identifying the latest address and contact information of a debtor who has lost contact, using resident registration databases or telecommunications company APIs.

[0634] "Means for generating documents to support legal procedures" refers to systems and technologies that automatically create documents necessary for legal responses such as arbitration procedures and bankruptcy procedures.

[0635] "Means for collecting the results of prompting activities and updating the model" refers to methods and techniques for evaluating the effectiveness of prompting activities, improving the analytical model based on that data, and increasing the accuracy of future prompting plans.

[0636] "Measures to prevent payment delays" refers to methods and technologies that use machine learning models to predict payment trends and take appropriate action in advance if there is a high possibility of a delay.

[0637] "Means for generating and sending notification messages" refers to systems or technologies for automatically creating and sending payment reminder messages to debtors at specified dates and times.

[0638] The following describes an embodiment of the present invention.

[0639] The server first collects data such as the debtor's payment history, communication history, and financial situation. In this step, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the communication history is obtained from a telecommunications company's API.

[0640] The server then analyzes the collected data. During data analysis, the data is preprocessed to remove noise and fill in missing values. After that, a machine learning algorithm is used to predict payment trends and the optimal timing for reminders. Specifically, past delay patterns are input into the model to predict the likelihood of the next delay.

[0641] Based on the analysis results, the server generates a customized reminder plan for each debtor. This plan includes the optimal reminder method (SNS notification, app notification, email) and the optimal timing. For example, if past data shows that a particular debtor is likely to respond to SNS notifications, the server will create a plan to first remind that debtor via SNS notification.

[0642] Based on the generated reminder plan, the server automatically generates and sends notification messages at the specified date and time. For example, when a payment deadline approaches, a notification message is automatically sent to the debtor's smartphone. The server also records the notification opening rate and payment completion rate to evaluate the effectiveness of the notifications.

[0643] For debtors who have become unreachable, the server tracks their latest contact information, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact details, allowing for lost contact to be reestablished and effective collection efforts to continue.

[0644] In addition, the server automatically generates documents required for mediation procedures, bankruptcy procedures, etc. to assist with necessary legal procedures, and the generated documents are sent to relevant departments and lawyers.

[0645] The server collects the results of debt collection activities and updates the model to reflect them in the next collection plan. For example, it analyzes email open rates and phone response status, and improves the accuracy of the model through a feedback loop.

[0646] Hardware and software used

[0647] In this system, software is installed on the server to collect data via the credit information agency's API, the bank's transaction database, and the telecommunications company's API. Python libraries such as pandas and scikit-learn are used for data analysis, and smtplib is used to generate and send notification messages. RandomForestClassifier is used as the machine learning model.

[0648] Specific examples

[0649] For example, the payment history of a certain debtor A over the past year is collected, and a machine learning model is used to predict the likelihood of the next payment being late. Based on the analysis results, an SNS notification is selected for Debtor A, and a notification message is sent the day before the payment due date. This notification message is automatically generated, and the open rate and response status after sending are recorded. This data is reflected in the next reminder plan.

[0650] Prompt Sentence Examples

[0651] "Please predict the likelihood of User A's next payment being late based on their bank payment history, communication history, and financial situation over the past year. Based on the results, please provide a Python script that will generate an optimal reminder plan and automate the process of sending a notification message to User A."

[0652] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[0653] Step 1:

[0654] Data collection: The server collects data such as the debtor's payment history, communication history, and financial situation via the API of credit information agencies, bank transaction databases, telecommunications company APIs, etc. Specifically, for a specific debtor, the payment history for the past year is obtained from the bank database, and the communication history is obtained from the telecommunications company API.

[0655] Input: Debtor identification information (e.g., Debtor ID)

[0656] Output: Datasets such as payment history, communication history, and financial status

[0657] Step 2:

[0658] Data preprocessing: The server preprocesses the collected data, removing noise and imputing missing values. This preprocessing cleans the data for later analysis. For example, missing values ​​are imputed with the mean value and outliers are removed.

[0659] Input: Collected dataset

[0660] Output: A preprocessed and clean dataset

[0661] Step 3:

[0662] Data analysis: The server analyzes the preprocessed data. It uses the machine learning algorithm RandomForestClassifier to predict payment trends and the optimal timing for reminders. It also inputs past delay patterns into the model to predict the likelihood of the next delay.

[0663] Input: Preprocessed clean dataset

[0664] Output: Predictions regarding payment trends and likelihood of delays

[0665] Step 4:

[0666] Generation of reminder plan: Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, if it is determined that a particular debtor is likely to respond to SNS notifications, it will create a reminder plan for that debtor via SNS notifications.

[0667] Input: Prediction result

[0668] Output: Optimal reminder plan

[0669] Step 5:

[0670] Generation and sending of notification messages: The server automatically generates and sends notification messages at the specified date and time based on the generated reminder plan. For example, a reminder message can be sent to the debtor's smartphone the day before the payment deadline.

[0671] Input: Optimal reminder schedule

[0672] Output: Notification message sent

[0673] Step 6:

[0674] Contact information tracking: For debtors who have become unable to contact us, the server uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, thereby restoring lost contact.

[0675] Input: Uncontactable debtor information

[0676] Output: Latest contact information

[0677] Step 7:

[0678] Generation of documents to support legal procedures: The server automatically generates documents required for necessary legal procedures (e.g., arbitration procedures, bankruptcy procedures) and sends them to the relevant departments and lawyers.

[0679] Input: Debtor's Legal Information

[0680] Output: Auto-generated legal document

[0681] Step 8:

[0682] Model update: The server collects the results of the reminder activities and updates the analytical model based on them. This improves the accuracy of the next reminder plan. For example, the server records the opening rate of notification messages and the payment completion rate and reflects them in the model.

[0683] Input: Result data of reminder activity

[0684] Output: Updated analytical model

[0685] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[0686] 1. Data Collection

[0687] Subject: Server

[0688] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains the payment history for the past year from the bank database, and the communication history (phone and email records) from the telecommunications company's API.

[0689] 2. Data Analysis

[0690] Subject: Server

[0691] The server preprocesses the collected data, removing noise and filling in missing values. It then uses machine learning algorithms to predict payment trends and the optimal timing for reminders. The server inputs past delay patterns into the model and calculates the risk of the next delay.

[0692] 3. Generate a customized reminder plan

[0693] Subject: Server

[0694] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data indicates that a debtor is likely to respond to email reminders, the server will first plan to remind that debtor by email.

[0695] 4. Utilizing the Emotion Engine

[0696] Subject: Server

[0697] The server uses an emotion engine to recognize the user's (debtor's) emotions during collection activities. The emotion engine detects emotions through voice and text analysis and identifies emotional states such as positive, negative, and neutral. For example, during a phone call, the server analyzes emotions from the user's tone of voice and the content of their text replies.

[0698] 5. Execute automatic reminders

[0699] Subject: Server

[0700] The server automatically executes reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. It also makes reminder calls via an automated calling system and plays recorded messages as needed. The emotion information recognized by the emotion engine is used to adjust the reminder plan.

[0701] 6. Tracking Lost Contacts

[0702] Subject: Server

[0703] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest address and contact information using a resident registration database or a telecommunications company's API. For example, it accesses a resident registration database to obtain the debtor's latest address information.

[0704] 7. Legal assistance

[0705] Subject: Server

[0706] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for sending to the relevant departments and lawyers.

[0707] 8. Feedback and optimization

[0708] Subject: Server

[0709] The server collects the results of collection activities and records them in a database. Information such as the open rate of emails sent, response status of phone calls, and whether or not payments were made is saved as a log. In addition, emotional information recognized by the emotion engine is also collected, and the generative AI model is updated based on the analysis results. This feedback loop improves the accuracy of collection plans for future collections, enabling more efficient debt collection.

[0710] Examples:

[0711] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0712] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and allows for more efficient and faster debt collection activities while taking into account the emotional state of the user.

[0713] The processing flow will be explained below.

[0714] Step 1:

[0715] Subject: Server

[0716] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains Debtor A's payment history for the past year from the bank database, and collects Debtor A's phone and email history for the past six months through the telecommunications company's API.

[0717] Step 2:

[0718] Subject: Server

[0719] The server preprocesses the collected data, which includes cleaning the data (removing noise) and imputing missing values. For example, the server removes incomplete data rows, imputes missing values ​​with the mean or median, and converts non-standardized data into a standardized format.

[0720] Step 3:

[0721] Subject: Server

[0722] The server analyzes the preprocessed data and extracts payment trends for each borrower. This analysis uses machine learning models to detect patterns of late payments from borrowers based on their past payment history. For example, the server can identify borrowers who have frequently made late payments in the past and predict the risk of the next late payment.

[0723] Step 4:

[0724] Subject: Server

[0725] The server generates an optimal reminder plan for each debtor. The generated plan includes the optimal reminder method (telephone, email, written) and optimal timing. Specifically, if a debtor has responded well to email reminders in the past, the server generates a plan that prioritizes email reminders.

[0726] Step 5:

[0727] Subject: Server

[0728] The server uses an emotion engine to recognize the user's emotions during the prompting activity. The emotion engine detects emotions through voice analysis and text analysis. For example, during a phone prompting call, the server can determine the user's emotional state (positive, negative, neutral, etc.) from the tone and speed of the user's voice and the content of the message.

[0729] Step 6:

[0730] Subject: Server

[0731] The server adjusts the reminder plan in real time based on the user's emotional information recognized by the emotion engine. For example, if the user shows negative emotions, the server may soften the tone of the reminder or temporarily suspend the reminder.

[0732] Step 7:

[0733] Subject: Server

[0734] The server automatically carries out reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder phone calls via an automated calling system and plays recorded messages. For example, the server automatically generates and sends a reminder email to debtor A on the 15th, and if there is no response, it makes a reminder phone call.

[0735] Step 8:

[0736] Subject: Server

[0737] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest addresses and contact information using a resident registration database or a telecommunications company's API. For example, it accesses the resident registration database to obtain the latest address information of Debtor A.

[0738] Step 9:

[0739] Subject: Server

[0740] The server automatically generates documents to support the necessary legal procedures. Mediation procedure documents and bankruptcy procedure documents are automatically created based on templates and sent to the relevant departments and lawyers. For example, the server automatically generates mediation procedure documents for Debtor A and prepares them to be sent to the lawyer.

[0741] Step 10:

[0742] Subject: Server

[0743] The server collects the results of the reminder activities and records them in a database. It stores information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made as logs. It also collects emotional information recognized by the emotion engine.

[0744] Step 11:

[0745] Subject: Server

[0746] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and apply new patterns and trends to it.

[0747] Example 2

[0748] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0749] Conventional debt collection systems have the problem of being inefficient and time-consuming, requiring a lot of manual work. It is also difficult to implement collection activities that take into account the debtor's feelings, and there is a lack of appropriate timing and means for approaching debtors. Furthermore, tracking down debtors who have become unreachable and generating the necessary documents for legal procedures are often done manually, which is labor-intensive and time-consuming. There is a need for a system that can solve these problems and achieve efficient and effective debt collection.

[0750] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[0751] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial situation; means for preprocessing the collected data to remove noise and fill in missing values; means for analyzing the preprocessed data and generating an optimal collection plan for each debtor; means for recognizing the debtor's emotions based on the analysis results and generating a customized collection plan; means for automatically conducting loan collection activities based on the generated collection plan; means for tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of the collection activities and updating the generation AI model based on the analysis results. This automates the debt collection process and enables efficient and prompt debt collection activities that take into account the user's emotional state.

[0752] "Data collection tools" refers to hardware and software used to obtain data such as debtors' payment history, communication history, and financial status.

[0753] The "preprocessing means" refers to a software module that has the function of removing noise and filling in missing values ​​from collected data.

[0754] "Data Analysis Tools" means machine learning algorithms and related software used to analyze pre-processed data and predict debtor behavior patterns and optimal collection strategies.

[0755] The "demand plan generation means" refers to software and algorithms that determine the means and timing of demands customized for each debtor based on the analysis results and generate a plan.

[0756] "Emotion Recognition Means" means software modules and algorithms for recognizing the emotional state of a debtor through speech and text analysis.

[0757] "Automated collection measures" refers to software and hardware that automates loan collection activities using email and automated calling systems based on generated collection plans.

[0758] "Tracking means" refers to software and hardware that has the ability to obtain the latest contact information of debtors who have become unreachable by using resident registration databases or telecommunications company APIs.

[0759] "Document Generation Measures" means software modules and templates for automatically generating arbitration documents and bankruptcy documents to support necessary legal proceedings.

[0760] The "feedback collection means" is a software module that collects the results of prompting activities and records them as a log in a database.

[0761] "Generative AI model update means" refers to a means for updating the generative AI model based on the results of analysis and prompting activities, thereby improving the accuracy of future prompting plans.

[0762] This invention is a system for streamlining debt management and collection operations, collecting necessary information from multiple data sources, automating analysis and collection activities, and further performing emotional analysis, enabling flexible responses according to the emotional state of the debtor.

[0763] Data collection

[0764] The server collects data such as the debtor's payment history, communication history, and financial situation. Specifically, it obtains the necessary information using the credit bureau's API, the bank's transaction database, and the telecommunications company's API. The server sends requests to these data sources, for example, accessing the "credit bureau API" endpoint to obtain the debtor's payment history. Similarly, it uses an SQL query to obtain transaction history for the past year from the "bank transaction database," and obtains phone and email records via the telecommunications company's API. This data is then stored in a "data collection folder" by the server.

[0765] Data analysis and preprocessing

[0766] The server preprocesses the collected data, removing noise and filling in missing values. It uses a "data cleaning function" to fill in missing values ​​and applies a "noise removal algorithm" to refine the data. After this process is complete, the preprocessed data is input into a "machine learning model" to predict payment trends and delay risk. Specifically, it uses a "payment pattern identification model" to analyze past payment data and calculate the risk of the next delay.

[0767] Generate a customized reminder plan

[0768] Based on the analysis results, the server generates the optimal reminder plan for each debtor. To generate the plan, a "plan generation algorithm" is used to determine the optimal reminder method (telephone, email, written) and timing. For example, in the case of "Debtor A," if past data shows that the debtor has a high response rate to email reminders, email is selected as the top priority method and entered into the reminder plan table.

[0769] Utilizing the Emotion Engine

[0770] The server uses an "emotion engine" to recognize the debtor's emotions during collection activities. This engine detects emotions by applying a "voice analysis module" to phone recordings and a "text analysis module" to email and chat content. Emotional states (positive, negative, neutral) are recorded in an "emotion log" to help adjust collection plans.

[0771] Execute automatic reminders

[0772] The server automatically executes reminder activities based on the generated reminder plan. The "reminder email generation module" generates the email content and sends it at the specified date and time using the "email sending script." It also makes reminder calls via the "automatic call system" and plays back recorded messages. The reminder plan is adjusted in real time using emotional information recognized by the emotion engine.

[0773] Tracking lost contacts

[0774] The server tracks the latest contact information of unreachable debtors using the resident registration database and the API of the telecommunications company. Specifically, it sends a request to the "resident registration API" to obtain the latest address and contact information. This information is then updated in the "contact database."

[0775] Legal process assistance

[0776] The server automatically generates documents to support the necessary legal procedures. Mediation and bankruptcy documents are created based on the "document generation template" and formatted by the "PDF generation module." The documents are then ready to be sent to the relevant departments and lawyers.

[0777] Feedback and Optimization

[0778] The server collects the results of the reminder activities and records them in a "feedback database." For example, it stores logs of the open rate of emails sent, response status of phone calls, and whether or not payments were made. It also collects emotional information recognized by the emotion engine and updates the "generative AI model." This feedback loop improves the accuracy of future reminder plans.

[0779] Examples:

[0780] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0781] Example prompt sentence:

[0782] "Calculate the optimal method and timing for reminding Debtor A based on their payment history and communication history over the past year. Then, optimize it using an emotion engine."

[0783] The above is a specific embodiment for carrying out the invention. This system automates the debt collection process and enables efficient and fast debt collection activities that take into account the emotional state of the user.

[0784] The flow of the identification process in the second embodiment will be described with reference to FIG.

[0785] Step 1:

[0786] The server collects data such as the debtor's payment history, communication history, and financial situation. The API of the credit information agency, the bank's transaction database, and the API endpoint of the telecommunications company are used as input. The server then sends requests to each API, obtains the required data, and saves it in a data collection folder. Specifically, it sends an API request, converts the returned data into a specified format, and saves it.

[0787] Step 2:

[0788] The server preprocesses the collected data. The input is the various collected data, and based on that, it performs noise removal and missing value imputation. It uses data cleaning functions to properly impute missing values ​​and applies noise removal algorithms. The output of this step is preprocessed data, which is sent to the next analysis step. Specific operations include filtering outliers and imputing missing values ​​using statistical methods.

[0789] Step 3:

[0790] The server analyzes the preprocessed data. The input is the preprocessed data, and the payment trends and late payment risk are analyzed based on a machine learning model. The received data is analyzed using a payment pattern identification model to calculate the late payment risk for each debtor. The results are output as an analysis report and used in the next step. Specifically, past payment data is input into the model to calculate the predicted late payment risk.

[0791] Step 4:

[0792] The server generates a customized reminder plan based on the analysis results. The input is the analysis report, and the plan generation algorithm is used to determine the individual reminder methods and timing. The generated plan is saved in the reminder plan table and sent to the execution step. Specifically, the server sets the optimal approach method and timing for each debtor based on the analysis results obtained from the database.

[0793] Step 5:

[0794] The server automatically executes reminder activities based on the generated reminder schedule. The input is data from the reminder schedule table, and the reminder email generation module generates the email content and sends it at the date and time specified by the email sending script. It also executes telephone reminders via an automatic calling system and plays back recorded messages. Specifically, it automatically executes the email script and sends an emergency message via the telephone system.

[0795] Step 6:

[0796] The server tracks the latest contact information for unreachable debtors. The input is the debtor information to be tracked, and the latest address and contact information is obtained using the resident registration database and the API of the telecommunications company. The server then updates the contact information in the contact database with the new information. Specifically, it sends a request to the resident registration API and updates the database with the obtained information.

[0797] Step 7:

[0798] The server automatically generates documents to support the necessary legal procedures. The input is the data required for the legal procedures, and mediation procedure documents and bankruptcy procedure documents are created using document generation templates and formatted in the PDF generation module. The generated documents are sent to relevant departments and lawyers. Specifically, the required information is entered into the document template, which is then automatically generated in PDF format.

[0799] Step 8:

[0800] The server collects the results of the prompting activities and updates the model. The inputs are the results of the prompting activities (email open rate, phone response status, etc.) and emotion recognition results, and the logs are saved in a feedback database. Based on this, the generative AI model is updated to improve the accuracy of future prompting plans. Specifically, the analysis results are added to the model learning data, and the model is retrained to improve accuracy.

[0801] The above are the specific processing steps of this system.

[0802] (Application example 2)

[0803] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the smart glasses 214 will be referred to as a "terminal."

[0804] In today's world, effective debt collection activities through communication with debtors are important, but traditional methods fail to fully consider the individual circumstances and emotions of debtors. Furthermore, the advertising industry faces challenges in accurately capturing users' interests and purchasing intent and delivering advertisements at the optimal time. To solve these challenges, a system is needed that analyzes the behavior and emotions of debtors and users and automatically responds optimally.

[0805] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[0806] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for analyzing the collected data to generate an optimal collection schedule for each debtor; means for automatically carrying out collection activities for debtors based on the generated collection schedule; means for searching for and tracking the latest contact information for debtors who have become unreachable; means for generating documents to support necessary legal procedures; means for collecting the results of the collection activities and updating a model based on the analysis results; means for collecting user purchase history, browsing history, and communication history; means for preprocessing the collected data and analyzing user interests and purchasing intentions using a machine learning algorithm; means for generating an optimal advertising plan for each user based on the analysis results; means for analyzing user emotions using an emotion analysis engine and adjusting advertisements; means for automatically delivering advertisements to users based on the generated advertising plan; and means for collecting the results of advertisement delivery and optimizing subsequent advertising plans. This automates the debt collection process and advertisement delivery process, enabling more efficient and effective responses while taking into account the user's emotional state and behavioral patterns.

[0807] A "debtor" is a person who has borrowed money from a financial institution or money lender and is obligated to repay the loan.

[0808] "Payment history" refers to a record of the payments a debtor has made to date, including the date, time, amount, and method of each payment.

[0809] "Communication history" refers to a record of contacts and conversations made through the means of communication used by the debtor (such as telephone or email).

[0810] "Financial situation" refers to the debtor's financial condition, including income, expenses, assets, and liabilities.

[0811] "Data collection methods" refer to the methods and techniques used to compile the required information, including APIs and databases.

[0812] "Data analysis tools" refer to techniques and methods used to pre-process and subsequently analyze collected data, including, for example, noise reduction and machine learning algorithms.

[0813] A "repayment plan" refers to specific methods and schedules for encouraging debtors to repay.

[0814] "Automatic collection activity means" refers to a technique or method for automatically sending collections to debtors based on the generated collection plan.

[0815] "Contact information retrieval means" refers to a technique or method for identifying the most current contact information for a debtor who has become unreachable.

[0816] "Document generation means for supporting legal procedures" refers to techniques and methods for automatically creating documents to be used in necessary legal procedures.

[0817] "Model update methods" refer to techniques and methods for updating generative algorithms and machine learning models based on analysis results, improving accuracy from the next time onwards.

[0818] "Purchase history" refers to a list or record of products a user has purchased in the past.

[0819] "Browsing history" refers to a record of the web pages and content a user has viewed on the Internet.

[0820] A "machine learning algorithm" is a mathematical model or computational procedure used to make predictions or classifications based on collected data.

[0821] An "advertising plan" refers to a specific method and schedule for delivering optimal advertisements to users.

[0822] An "emotion analysis engine" is software or algorithms for detecting and analyzing a user's emotional state.

[0823] "Automatic advertisement distribution means" refers to a technique or method for automatically distributing advertisements to users in accordance with a generated advertising plan.

[0824] "Advertising delivery result optimization means" refers to techniques and methods for optimizing future advertising plans based on the results of advertising delivery.

[0825] System Overview

[0826] The system of the present invention automates the collection and analysis of data such as debtor payment history, communication history, and financial situation, and has the function of generating an optimal collection plan and automatically carrying out collection activities.It can also be applied to the advertising field, and has the function of delivering optimal advertisements based on users' purchase history, browsing history, and emotional information.The system consists of the following elements: server, terminal, and user.

[0827] Program Details

[0828] Hardware and software used

[0829] Hardware: Servers, smartphones, smart glasses, head-mounted displays

[0830] Software: Emotion engine (Emotion API), machine learning model (TensorFlow), database (MySQL), ad delivery management software (Ad Manager API)

[0831] Data collection

[0832] The server uses APIs and databases to collect data on the debtor's payment history, communication history, and financial status. For example, it obtains payment history for the past year from a bank database and call history from a telecommunications company's API.

[0833] Data analysis

[0834] The server preprocesses the collected data, removing noise and filling in missing values, and then analyzes the data using a machine learning algorithm (using TensorFlow) to predict the debtor's payment trends and the optimal timing for collection.

[0835] Generate reminder schedule

[0836] Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, for debtors for whom email reminders are effective, email reminders are sent first, and if there is no response, the server switches to phone reminders.

[0837] Utilizing the Emotion Engine

[0838] The server uses the emotion engine to analyze the debtor's emotions during collection activities, for example, by analyzing the debtor's tone of voice during a collection call, and if negative emotions are detected, adjust the collection plan accordingly.

[0839] Execute automatic reminders

[0840] Based on the generated reminder plan, the server automatically executes reminder activities, automatically generates reminder emails and sends them at the specified date and time, and also makes reminder calls via an automated calling system and plays recorded messages.

[0841] Collection of user purchasing and browsing history

[0842] The server collects the user's purchase history and browsing history using various APIs, such as the History API of a web browser or the API of a communication app.

[0843] Advertising plan generation and ad delivery

[0844] The server preprocesses the collected user data and uses machine learning algorithms to analyze the user's interests and purchasing intent. Based on the analysis results, it generates an optimal advertising plan for each user. Based on the generated advertising plan, the server automatically delivers advertisements to the user.

[0845] Result feedback and model optimization

[0846] The server records the results of ad delivery in a database and optimizes future ad plans. Ad click rates and purchase rates are fed back into the model to improve the ad delivery algorithm.

[0847] Specific examples

[0848] The server retrieves debtor A's payment history for the past year from the bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th of each month, and if there is no response, a reminder phone call is made. Similarly, the server analyzes user B's past purchase and browsing history and delivers the optimal advertisement during the lunch break.

[0849] Prompt Sentence Examples

[0850] "We analyze a user's purchasing and browsing history over the past year to predict what products they are likely to purchase next. We use the Emotion API to analyze their emotions while watching an ad, and then display ads from categories that they responded positively to."

[0851] This will automate the debt collection and advertising delivery processes, enabling more efficient and effective operations.

[0852] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[0853] Step 1:

[0854] Data collection

[0855] The server uses APIs and databases to collect the necessary data.

[0856] Input: Bank database API, Telecommunications company API, Web browser History API, Telecommunications app API

[0857] Specific operation: Obtain the debtor's payment history for the past year from the bank database, obtain call history from the telecommunications company's API, and obtain the user's purchase history and browsing history using the web browser's History API and the communication app's API.

[0858] Output: Data set of payment history, communication history, purchase history, browsing history, etc. for the relevant debtors and users

[0859] Step 2:

[0860] Data Preprocessing

[0861] The server preprocesses the collected data, removing noise and completing missing values.

[0862] Input: The dataset collected in step 1

[0863] Specific actions: Identify and remove or correct inaccurate information in collected data, and impute missing values ​​in an appropriate manner.

[0864] Output: A clean and reliable dataset

[0865] Step 3:

[0866] Data analysis

[0867] The server analyzes the pre-processed data using machine learning algorithms to predict the debtor's payment trends and the optimal timing for reminders.

[0868] Input: The dataset preprocessed in step 2, the machine learning algorithm (TensorFlow)

[0869] How it works: It uses machine learning algorithms to analyze each debtor's payment habits and risk of late payments, and also predicts the next likely purchase based on the user's purchase and browsing history.

[0870] Output: Analysis results on payment trends, optimal reminder timing, user interests, and purchasing intent

[0871] Step 4:

[0872] Generate promotional and advertising plans

[0873] Based on the analysis results, the server generates an optimal reminder plan for each debtor and an optimal advertising plan for each user.

[0874] Input: Analysis results obtained in Step 3

[0875] Specific operation: Based on the analysis results, the system determines the optimal reminder method (telephone, email, written) and timing for each debtor and generates a reminder plan. It also determines the optimal advertising content and distribution timing for each user and generates an advertising plan.

[0876] Output: Reminder plan for each debtor, advertising plan for each user

[0877] Step 5:

[0878] Emotion analysis

[0879] The server uses an emotion engine to analyze the user's emotions during the promotional activities and advertisement viewing.

[0880] Input: Voice data during promotional activities and ad viewing, text data, emotion engine (Emotion API)

[0881] Specific operation: The emotion engine analyzes voice and text data to identify emotional states such as positive, negative, or neutral.

[0882] Output: Analysis results on user sentiment

[0883] Step 6:

[0884] Execution of automatic reminders and advertisement distribution

[0885] The server automatically executes promotional activities and advertisement distribution based on the generated plan.

[0886] Input: Promotion plan and advertising plan generated in step 4

[0887] Specific operations: Automatically generate and send reminder emails to debtors at the specified date and time. Also, make reminder phone calls via an automated calling system. Furthermore, distribute advertisements at the specified times based on the advertising plan.

[0888] Output: Reminders performed and ads delivered

[0889] Step 7:

[0890] Result feedback and optimization

[0891] The server collects the results of the promotional activities and advertisement distribution and optimizes the plan for the next time.

[0892] Input: Promotional activity result data, ad delivery result data

[0893] Specific operations: Results such as the open rate of reminder emails, response status of phone calls, click rates and purchase rates of ads are recorded in a database, and the model is retrained and optimized.

[0894] Output: Updated machine learning models and optimization data for next promotional and advertising plans

[0895] Through these steps, the system of the invention automates the debt collection and advertising distribution processes, achieving effective and efficient operations.

[0896] The specific processing unit 290 transmits the result of the specific processing to the smart glasses 214. In the smart glasses 214, the control unit 46A causes the speaker 240 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[0897] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0898] In the above embodiment, an example in which the specific processing is performed by the data processing device 12 has been given, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the smart glasses 214.

[0899] [Third embodiment]

[0900] FIG. 5 shows an example of the configuration of a data processing system 310 according to the third embodiment.

[0901] 5, the data processing system 310 includes the data processing device 12 and a headset type terminal 314. An example of the data processing device 12 is a server.

[0902] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0903] The headset type terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a display 343. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the display 343 are also connected to the bus 52.

[0904] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[0905] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[0906] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[0907] Fig. 6 shows an example of the main functions of the data processing device 12 and the headset type terminal 314. As shown in Fig. 6, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[0908] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[0909] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0910] In the headset type terminal 314, a reception output process is performed by the processor 46. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[0911] Next, a description will be given of the identification process performed by the identification processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as the "server" and the headset type terminal 314 will be referred to as the "terminal."

[0912] 1. Data Collection

[0913] Subject: Server

[0914] The server first collects data such as the debtor's payment history, communication history, and financial situation. At this stage, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the email communication history is obtained from the email server.

[0915] 2. Data Analysis

[0916] Subject: Server

[0917] The server analyzes the collected data. When analyzing the data, it first preprocesses it to remove noise and fill in missing values. It then uses a machine learning algorithm to predict payment trends and the optimal timing for reminders. Specifically, it inputs past delay patterns into a model to predict the likelihood of the next delay.

[0918] 3. Generate a customized reminder plan

[0919] Subject: Server

[0920] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will create a plan to first remind that debtor by email.

[0921] 4. Execute automatic reminders

[0922] Subject: Server

[0923] The server automatically executes reminder activities based on the generated reminder plan. For example, it automatically generates and sends reminder emails at a specified date and time. If a reminder call is required, it makes a call via an automated calling system and plays a recorded message.

[0924] 5. Tracking Lost Contacts

[0925] Subject: Server

[0926] The server tracks the latest contact information of debtors who have become unreachable, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, allowing lost contact to be restored.

[0927] 6. Legal assistance

[0928] Subject: Server

[0929] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[0930] 7. Feedback and optimization

[0931] Subject: Server

[0932] The server collects the results of collection activities and updates the model. For example, it records the open rate of emails and the response status of phone calls, and reflects this in the next collection plan. This feedback loop improves the accuracy of future collection plans, enabling more efficient debt collection.

[0933] Examples:

[0934] For example, the server retrieves debtor A's payment history for the past year via the bank database and API, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a further reminder phone call is made. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[0935] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and enables more efficient and faster debt collection activities.

[0936] The processing flow will be explained below.

[0937] Step 1:

[0938] Subject: Server

[0939] The server collects data such as the debtor's payment history, communication history, and financial situation. This collection process obtains the necessary information using the API of a credit information agency, a bank's transaction database, and a telecommunications company's API. For example, the server sends a query to a bank database to obtain a specific debtor's payment history for the past year. It also obtains communication history (phone and email records) from a telecommunications company's API.

[0940] Step 2:

[0941] Subject: Server

[0942] The server preprocesses the collected data, including cleaning the data (removing noise) and imputing missing values. Specific operations include deleting incomplete data rows and imputing missing values ​​with the mean or median. It also converts data in non-standard formats into a standard format.

[0943] Step 3:

[0944] Subject: Server

[0945] The server analyzes the preprocessed data and extracts payment trends for each debtor. This analysis uses machine learning models to detect debtor delay patterns from past payment history. For example, the server can identify debtors who have had many late payments in the past and calculate the risk of their next payment delay.

[0946] Step 4:

[0947] Subject: Server

[0948] The server generates an optimal reminder plan for each debtor. The generated reminder plan includes the optimal reminder method (phone, email, written) and optimal timing. For example, if a debtor has responded well to emails in the past, the server generates a plan that prioritizes email reminders.

[0949] Step 5:

[0950] Subject: Server

[0951] The server automatically executes reminder activities based on the generated reminder plan. Specifically, it automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder calls via an automated calling system and plays recorded messages.

[0952] Step 6:

[0953] Subject: Server

[0954] The server tracks the latest contact information of debtors who have become unreachable. To do this, it uses a resident registration database or a telecommunications company's API to identify the latest address and contact information. For example, the server accesses a resident registration database to obtain the debtor's latest address information.

[0955] Step 7:

[0956] Subject: Server

[0957] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for delivery to relevant departments and lawyers.

[0958] Step 8:

[0959] Subject: Server

[0960] The server collects the results of the collection activities and records them in a database, for example, by saving information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made.

[0961] Step 9:

[0962] Subject: Server

[0963] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and incorporate new patterns and trends.

[0964] Example 1

[0965] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[0966] In conventional debt collection systems, collection plans for each debtor are general and uniform, and collections are not optimally tailored to the circumstances of each debtor. This reduces collection efficiency and makes it difficult to track debtors when they cannot be contacted. Furthermore, preparing documents required for legal procedures is a manual process that consumes time and resources.

[0967] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[0968] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for preprocessing the collected data, removing noise, and completing missing values; means for analyzing the data using a machine learning algorithm and predicting the optimal timing and means of collection for each debtor; means for generating a collection plan customized for each debtor based on the prediction; means for automatically carrying out collection activities for debtors based on the generated collection plan; means for searching for and tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of collection activities and updating the model based on the analysis results. This automates the debt collection process, enables optimal collection activities tailored to the circumstances of individual debtors, efficient tracking when debtors cannot be contacted, and enables the rapid and accurate preparation of legal procedure documents.

[0969] A "debtor" is an individual or legal entity that has a financial obligation in the form of a debt, loan, or other form of debt.

[0970] "Payment history" is a record of past payments made by a debtor, including details such as due dates, amounts, and payment methods.

[0971] "Communication history" refers to a record of communication activities such as phone calls, emails, and messages between you and the debtor.

[0972] "Financial status" refers to information that indicates the debtor's current financial condition, and includes data on income, expenses, assets, liabilities, and the like.

[0973] "Data collection means" refers to a method or device by which the server obtains information such as the debtor's payment history, communication history, and financial situation.

[0974] "Data preprocessing" is the process of preparing collected data before analysis, such as removing noise from the data and filling in missing values.

[0975] "Machine learning algorithms" are statistical methods and models for finding patterns and knowledge in data.

[0976] An "analysis means" is a method or device for predicting a debtor's behavioral tendencies and the optimal timing for making a collection request by utilizing collected and preprocessed data.

[0977] A "demand plan" is a plan that determines the optimal means and timing of demands to prevent a debtor from delaying repayments.

[0978] A "customized reminder plan" is a reminder plan that is individually tailored based on each debtor's characteristics and behavior.

[0979] The "automatic prompting activity execution means" is a method or device that allows the server to automatically execute prompting means such as telephone calls or e-mails based on the generated prompting plan.

[0980] A "contact information tracking means" is a method or device for searching for and tracking the latest contact information of a debtor who has become unreachable.

[0981] "Legal procedure support means" refers to a method or device for generating documents related to necessary legal procedures and providing them to relevant departments and experts.

[0982] The "feedback collection means" is a method or device for collecting the results of prompting activities and reflecting them in the next prompting plan.

[0983] A "model updater" is a method or device for retraining an analytical model based on collected feedback data to improve its accuracy.

[0984] The present invention provides a system for automating and efficiently carrying out debt collection activities against specific debtors. A specific embodiment of this system will now be described.

[0985] Hardware and software used

[0986] This system operates in a server-based network environment. The specific hardware and software that supports each process in this system are listed below.

[0987] Server: Collects data, performs analytics, performs remediation activities, tracks contact information, assists with legal proceedings, gathers feedback, and updates the model.

[0988] Example of hardware used: A server equipped with a high-performance CPU and large-capacity RAM

[0989] Examples of software used: MongoDB (database), Pandas (data processing), Scikit-learn (machine learning), Twilio (automated calls), SMTP server (email sending)

[0990] Terminal: A device that interacts with a server and displays or inputs information.

[0991] Examples of hardware used: personal computers, tablets, smartphones

[0992] Specific processing of the system

[0993] Data collection

[0994] The server uses APIs from credit information agencies, banks, and telecommunications companies to collect data such as debtor payment history, communication history, and financial status. For example, the server obtains debtor A's payment history for the past year from the bank API and stores that data in MongoDB.

[0995] Data Preprocessing

[0996] The collected data is first preprocessed. The server uses the Pandas library to remove noise and impute missing values, then formats the collected data into a data frame.

[0997] Data analysis

[0998] Machine learning algorithms are applied to the preprocessed data. The server uses Scikit-learn and models such as random forests to predict the debtor's payment trends and the optimal timing for reminders. For example, it analyzes the debtor's delay patterns and predicts the likelihood of the next delay.

[0999] Generate a customized reminder plan

[1000] Based on the analysis results, the server generates a customized reminder plan for each debtor. This reminder plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will plan to first remind that debtor by email.

[1001] Execute automatic reminders

[1002] Based on the generated reminder plan, the server automatically executes reminder activities. It automatically generates reminder emails at the specified date and time and sends them via the SMTP server. In addition, if a phone reminder is required, it makes an automated call via Twilio's API and plays a pre-recorded message.

[1003] Tracking lost contacts

[1004] For debtors who can no longer be contacted, the server uses the resident registration database or the telecommunications company's API to search for the latest contact information and update the database.

[1005] Legal process assistance

[1006] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[1007] Gathering feedback and updating the model

[1008] The server records the results of each reminder activity (e.g., email open rates and phone call response rates) and collects them as feedback data. The collected feedback data is used to retrain the analysis model and improve the accuracy of the next reminder plan.

[1009] Specific examples

[1010] For example, the server retrieves debtor A's payment history for the past year from the bank API and stores the data in JSON format in MongoDB. Next, it uses the Pandas library to fill in missing values ​​in the data and uses Scikit-learn to predict payment trends using a random forest model. Based on the analysis results, it determines the optimal method of reminding debtor A, automatically generates and sends a reminder email on the 15th. If there is no response, it makes a reminder phone call using Twilio's API. The results are recorded in the database and reflected in the next model update.

[1011] Prompt Sentence Examples

[1012] "Based on Debtor A's payment history, email communication history, and phone call history for the past year, please predict the risk of delay in the next debt payment and plan the optimal method and timing of reminder payments."

[1013] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1014] Program processing steps

[1015] Step 1: Data collection

[1016] Subject: Server

[1017] The server receives the debtor's identification information as input. The server collects data such as the debtor's payment history, communication history, and financial status through the APIs of credit bureaus, banks, and telecommunications companies. For example, the server obtains the payment history of debtor A through an API call and saves the data in JSON format. During this process, the server stores the obtained data in a database.

[1018] Input: Debtor Identification Information

[1019] Output: Data on debtor's payment history, communication history, financial status, etc. (JSON format)

[1020] Step 2: Data Preprocessing

[1021] Subject: Server

[1022] The server receives the collected data as input. It uses the Pandas library to preprocess the data, removing noise, imputing missing values, and preparing it in a format suitable for analysis. During this process, it generates a data frame and performs operations such as removing outliers.

[1023] Input: Collected data (JSON format)

[1024] Output: Preprocessed data (data frame)

[1025] Step 3: Data analysis

[1026] Subject: Server

[1027] The server receives preprocessed data as input. It uses Scikit-learn to apply machine learning algorithms (e.g., random forests) to predict payment trends and optimal reminder timing. The analysis results are output as predictions for each debtor, including the likelihood of the next late payment and the optimal reminder method.

[1028] Input: Preprocessed data (data frame)

[1029] Output: Analysis results (payment trends, optimal reminder timing)

[1030] Step 4: Generate a customized reminder plan

[1031] Subject: Server

[1032] The server receives the analysis results as input. Based on this, it generates a customized reminder plan for each debtor. For example, if the analysis shows that email reminders are effective for debtor A, the server determines the content and timing of the emails.

[1033] Input: Analysis results (payment trends, optimal reminder timing)

[1034] Output: Customized reminder plan

[1035] Step 5: Run automatic reminders

[1036] Subject: Server

[1037] The server takes your customized reminder plan as input and automatically executes reminder activities at the specified dates and times, for example, generating and sending reminder emails using an SMTP server, and optionally placing automated calls using Twilio's APIs to play pre-recorded messages.

[1038] Input: Customized reminder plan

[1039] Output: Reminder actions taken (emails sent, call logs)

[1040] Step 6: Track down the missing person

[1041] Subject: Server

[1042] The server receives contact information for unreachable debtors as input, and uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information and update the database.

[1043] Input: Unreachable person's contact information

[1044] Output: Latest updated contact information

[1045] Step 7: Legal assistance

[1046] Subject: Server

[1047] The server receives the necessary legal procedure information as input, and then automatically generates mediation and bankruptcy procedure documents based on templates and sends them to the relevant departments and lawyers.

[1048] Input: Legal process information and templates

[1049] Output: Auto-generated legal document

[1050] Step 8: Gather feedback and update the model

[1051] Subject: Server

[1052] The server receives as input the results of each reminder activity (e.g., email open rates and phone response rates), which are collected as feedback data and used to retrain the analytical model, improving the accuracy of the next reminder plan.

[1053] Input: Results of reminder activities (open rate, response status)

[1054] Output: Updated analytical model

[1055] These are the specific processing steps of this system's program. Through this process, collection activities can be optimized for each debtor, greatly improving the efficiency of debt collection.

[1056] (Application example 1)

[1057] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1058] In the debt collection process, debtors frequently experience payment delays and are unable to be contacted, resulting in a decline in collection efficiency. Other issues include the time and cost required to simultaneously contact multiple debtors and the preparation of legal paperwork. Furthermore, it can be difficult to determine the optimal timing for collection, which can result in the ineffectiveness of payment collection efforts. These issues need to be resolved.

[1059] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1060] In this invention, the server includes means for collecting data such as debtor payment history, communication history, and financial status, means for analyzing the collected data to generate an optimal collection plan for each debtor, means for automatically carrying out collection activities for debtors based on the generated collection plan, means for searching for and tracking the latest contact information for debtors who have become unreachable, means for generating documents to support necessary legal procedures, means for collecting the results of the collection activities and updating the model based on the analysis results, means for predicting payment trends and determining the optimal timing for collection activities to prevent payment delays, and means for automatically generating and sending notification messages at specified dates and times. This streamlines the debt collection process and enables fast and effective debt collection activities.

[1061] A "debtor" is an individual or corporation that is obligated to borrow money from a financial institution, business partner, etc., or to pay for goods or services.

[1062] A "payment history" is a record of past payments made by a debtor, including details such as the date, amount, and method of payment within a specific period of time.

[1063] "Communication history" refers to a record of the debtor's communication activities, such as phone calls, emails, and messages, including the date, time, frequency, and content of the messages sent and received.

[1064] "Financial situation" refers to information that shows the debtor's economic status, such as income, expenses, assets, and liabilities.

[1065] "Means of collecting data" refers to the mechanisms and technologies for collecting the necessary data using APIs from credit information agencies, bank transaction databases, telecommunications company APIs, etc.

[1066] "Means for analyzing data" refers to methods and techniques for preprocessing collected data and using machine learning models to predict payment trends and find the optimal timing for reminders.

[1067] A "demand collection plan" is a plan that specifically outlines when and how to request payment from each debtor.

[1068] "Means for carrying out collection activities" refers to systems or technologies that automatically notify or contact debtors based on the generated collection plan.

[1069] "Means for searching and tracking contact information" refers to methods and technologies for identifying the latest address and contact information of a debtor who has lost contact, using resident registration databases or telecommunications company APIs.

[1070] "Means for generating documents to support legal procedures" refers to systems and technologies that automatically create documents necessary for legal responses such as arbitration procedures and bankruptcy procedures.

[1071] "Means for collecting the results of prompting activities and updating the model" refers to methods and techniques for evaluating the effectiveness of prompting activities, improving the analytical model based on that data, and increasing the accuracy of future prompting plans.

[1072] "Measures to prevent payment delays" refers to methods and technologies that use machine learning models to predict payment trends and take appropriate action in advance if there is a high possibility of a delay.

[1073] "Means for generating and sending notification messages" refers to systems or technologies for automatically creating and sending payment reminder messages to debtors at specified dates and times.

[1074] The following describes an embodiment of the present invention.

[1075] The server first collects data such as the debtor's payment history, communication history, and financial situation. In this step, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the communication history is obtained from a telecommunications company's API.

[1076] The server then analyzes the collected data. During data analysis, the data is preprocessed to remove noise and fill in missing values. After that, a machine learning algorithm is used to predict payment trends and the optimal timing for reminders. Specifically, past delay patterns are input into the model to predict the likelihood of the next delay.

[1077] Based on the analysis results, the server generates a customized reminder plan for each debtor. This plan includes the optimal reminder method (SNS notification, app notification, email) and the optimal timing. For example, if past data shows that a particular debtor is likely to respond to SNS notifications, the server will create a plan to first remind that debtor via SNS notification.

[1078] Based on the generated reminder plan, the server automatically generates and sends notification messages at the specified date and time. For example, when a payment deadline approaches, a notification message is automatically sent to the debtor's smartphone. The server also records the notification opening rate and payment completion rate to evaluate the effectiveness of the notifications.

[1079] For debtors who have become unreachable, the server tracks their latest contact information, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact details, allowing for lost contact to be reestablished and effective collection efforts to continue.

[1080] In addition, the server automatically generates documents required for mediation procedures, bankruptcy procedures, etc. to assist with necessary legal procedures, and the generated documents are sent to relevant departments and lawyers.

[1081] The server collects the results of debt collection activities and updates the model to reflect them in the next collection plan. For example, it analyzes email open rates and phone response status, and improves the accuracy of the model through a feedback loop.

[1082] Hardware and software used

[1083] In this system, software is installed on the server to collect data via the credit information agency's API, the bank's transaction database, and the telecommunications company's API. Python libraries such as pandas and scikit-learn are used for data analysis, and smtplib is used to generate and send notification messages. RandomForestClassifier is used as the machine learning model.

[1084] Specific examples

[1085] For example, the payment history of a certain debtor A over the past year is collected, and a machine learning model is used to predict the likelihood of the next payment being late. Based on the analysis results, an SNS notification is selected for Debtor A, and a notification message is sent the day before the payment due date. This notification message is automatically generated, and the open rate and response status after sending are recorded. This data is reflected in the next reminder plan.

[1086] Prompt Sentence Examples

[1087] "Please predict the likelihood of User A's next payment being late based on their bank payment history, communication history, and financial situation over the past year. Based on the results, please provide a Python script that will generate an optimal reminder plan and automate the process of sending a notification message to User A."

[1088] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1089] Step 1:

[1090] Data collection: The server collects data such as the debtor's payment history, communication history, and financial situation via the API of credit information agencies, bank transaction databases, telecommunications company APIs, etc. Specifically, for a specific debtor, the payment history for the past year is obtained from the bank database, and the communication history is obtained from the telecommunications company API.

[1091] Input: Debtor identification information (e.g., Debtor ID)

[1092] Output: Datasets such as payment history, communication history, and financial status

[1093] Step 2:

[1094] Data preprocessing: The server preprocesses the collected data, removing noise and imputing missing values. This preprocessing cleans the data for later analysis. For example, missing values ​​are imputed with the mean value and outliers are removed.

[1095] Input: Collected dataset

[1096] Output: A preprocessed and clean dataset

[1097] Step 3:

[1098] Data analysis: The server analyzes the preprocessed data. It uses the machine learning algorithm RandomForestClassifier to predict payment trends and the optimal timing for reminders. It also inputs past delay patterns into the model to predict the likelihood of the next delay.

[1099] Input: Preprocessed clean dataset

[1100] Output: Predictions regarding payment trends and likelihood of delays

[1101] Step 4:

[1102] Generation of reminder plan: Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, if it is determined that a particular debtor is likely to respond to SNS notifications, it will create a reminder plan for that debtor via SNS notifications.

[1103] Input: Prediction result

[1104] Output: Optimal reminder plan

[1105] Step 5:

[1106] Generation and sending of notification messages: The server automatically generates and sends notification messages at the specified date and time based on the generated reminder plan. For example, a reminder message can be sent to the debtor's smartphone the day before the payment deadline.

[1107] Input: Optimal reminder schedule

[1108] Output: Notification message sent

[1109] Step 6:

[1110] Contact information tracking: For debtors who have become unable to contact us, the server uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, thereby restoring lost contact.

[1111] Input: Uncontactable debtor information

[1112] Output: Latest contact information

[1113] Step 7:

[1114] Generation of documents to support legal procedures: The server automatically generates documents required for necessary legal procedures (e.g., arbitration procedures, bankruptcy procedures) and sends them to the relevant departments and lawyers.

[1115] Input: Debtor's Legal Information

[1116] Output: Auto-generated legal document

[1117] Step 8:

[1118] Model update: The server collects the results of the reminder activities and updates the analytical model based on them. This improves the accuracy of the next reminder plan. For example, the server records the opening rate of notification messages and the payment completion rate and reflects them in the model.

[1119] Input: Result data of reminder activity

[1120] Output: Updated analytical model

[1121] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1122] 1. Data Collection

[1123] Subject: Server

[1124] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains the payment history for the past year from the bank database, and the communication history (phone and email records) from the telecommunications company's API.

[1125] 2. Data Analysis

[1126] Subject: Server

[1127] The server preprocesses the collected data, removing noise and filling in missing values. It then uses machine learning algorithms to predict payment trends and the optimal timing for reminders. The server inputs past delay patterns into the model and calculates the risk of the next delay.

[1128] 3. Generate a customized reminder plan

[1129] Subject: Server

[1130] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data indicates that a debtor is likely to respond to email reminders, the server will first plan to remind that debtor by email.

[1131] 4. Utilizing the Emotion Engine

[1132] Subject: Server

[1133] The server uses an emotion engine to recognize the user's (debtor's) emotions during collection activities. The emotion engine detects emotions through voice and text analysis and identifies emotional states such as positive, negative, and neutral. For example, during a phone call, the server analyzes emotions from the user's tone of voice and the content of their text replies.

[1134] 5. Execute automatic reminders

[1135] Subject: Server

[1136] The server automatically executes reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. It also makes reminder calls via an automated calling system and plays recorded messages as needed. The emotion information recognized by the emotion engine is used to adjust the reminder plan.

[1137] 6. Tracking Lost Contacts

[1138] Subject: Server

[1139] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest address and contact information using a resident registration database or a telecommunications company's API. For example, it accesses a resident registration database to obtain the debtor's latest address information.

[1140] 7. Legal assistance

[1141] Subject: Server

[1142] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for sending to the relevant departments and lawyers.

[1143] 8. Feedback and optimization

[1144] Subject: Server

[1145] The server collects the results of collection activities and records them in a database. Information such as the open rate of emails sent, response status of phone calls, and whether or not payments were made is saved as a log. In addition, emotional information recognized by the emotion engine is also collected, and the generative AI model is updated based on the analysis results. This feedback loop improves the accuracy of collection plans for future collections, enabling more efficient debt collection.

[1146] Examples:

[1147] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[1148] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and allows for more efficient and faster debt collection activities while taking into account the emotional state of the user.

[1149] The processing flow will be explained below.

[1150] Step 1:

[1151] Subject: Server

[1152] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains Debtor A's payment history for the past year from the bank database, and collects Debtor A's phone and email history for the past six months through the telecommunications company's API.

[1153] Step 2:

[1154] Subject: Server

[1155] The server preprocesses the collected data, which includes cleaning the data (removing noise) and imputing missing values. For example, the server removes incomplete data rows, imputes missing values ​​with the mean or median, and converts non-standardized data into a standardized format.

[1156] Step 3:

[1157] Subject: Server

[1158] The server analyzes the preprocessed data and extracts payment trends for each borrower. This analysis uses machine learning models to detect patterns of late payments from borrowers based on their past payment history. For example, the server can identify borrowers who have frequently made late payments in the past and predict the risk of the next late payment.

[1159] Step 4:

[1160] Subject: Server

[1161] The server generates an optimal reminder plan for each debtor. The generated plan includes the optimal reminder method (telephone, email, written) and optimal timing. Specifically, if a debtor has responded well to email reminders in the past, the server generates a plan that prioritizes email reminders.

[1162] Step 5:

[1163] Subject: Server

[1164] The server uses an emotion engine to recognize the user's emotions during the prompting activity. The emotion engine detects emotions through voice analysis and text analysis. For example, during a phone prompting call, the server can determine the user's emotional state (positive, negative, neutral, etc.) from the tone and speed of the user's voice and the content of the message.

[1165] Step 6:

[1166] Subject: Server

[1167] The server adjusts the reminder plan in real time based on the user's emotional information recognized by the emotion engine. For example, if the user shows negative emotions, the server may soften the tone of the reminder or temporarily suspend the reminder.

[1168] Step 7:

[1169] Subject: Server

[1170] The server automatically carries out reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder phone calls via an automated calling system and plays recorded messages. For example, the server automatically generates and sends a reminder email to debtor A on the 15th, and if there is no response, it makes a reminder phone call.

[1171] Step 8:

[1172] Subject: Server

[1173] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest addresses and contact information using a resident registration database or a telecommunications company's API. For example, it accesses the resident registration database to obtain the latest address information of Debtor A.

[1174] Step 9:

[1175] Subject: Server

[1176] The server automatically generates documents to support the necessary legal procedures. Mediation procedure documents and bankruptcy procedure documents are automatically created based on templates and sent to the relevant departments and lawyers. For example, the server automatically generates mediation procedure documents for Debtor A and prepares them to be sent to the lawyer.

[1177] Step 10:

[1178] Subject: Server

[1179] The server collects the results of the reminder activities and records them in a database. It stores information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made as logs. It also collects emotional information recognized by the emotion engine.

[1180] Step 11:

[1181] Subject: Server

[1182] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and apply new patterns and trends to it.

[1183] Example 2

[1184] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1185] Conventional debt collection systems have the problem of being inefficient and time-consuming, requiring a lot of manual work. It is also difficult to implement collection activities that take into account the debtor's feelings, and there is a lack of appropriate timing and means for approaching debtors. Furthermore, tracking down debtors who have become unreachable and generating the necessary documents for legal procedures are often done manually, which is labor-intensive and time-consuming. There is a need for a system that can solve these problems and achieve efficient and effective debt collection.

[1186] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1187] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial situation; means for preprocessing the collected data to remove noise and fill in missing values; means for analyzing the preprocessed data and generating an optimal collection plan for each debtor; means for recognizing the debtor's emotions based on the analysis results and generating a customized collection plan; means for automatically conducting loan collection activities based on the generated collection plan; means for tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of the collection activities and updating the generation AI model based on the analysis results. This automates the debt collection process and enables efficient and prompt debt collection activities that take into account the user's emotional state.

[1188] "Data collection tools" refers to hardware and software used to obtain data such as debtors' payment history, communication history, and financial status.

[1189] The "preprocessing means" refers to a software module that has the function of removing noise and filling in missing values ​​from collected data.

[1190] "Data Analysis Tools" means machine learning algorithms and related software used to analyze pre-processed data and predict debtor behavior patterns and optimal collection strategies.

[1191] The "demand plan generation means" refers to software and algorithms that determine the means and timing of demands customized for each debtor based on the analysis results and generate a plan.

[1192] "Emotion Recognition Means" means software modules and algorithms for recognizing the emotional state of a debtor through speech and text analysis.

[1193] "Automated collection measures" refers to software and hardware that automates loan collection activities using email and automated calling systems based on generated collection plans.

[1194] "Tracking means" refers to software and hardware that has the ability to obtain the latest contact information of debtors who have become unreachable by using resident registration databases or telecommunications company APIs.

[1195] "Document Generation Measures" means software modules and templates for automatically generating arbitration documents and bankruptcy documents to support necessary legal proceedings.

[1196] The "feedback collection means" is a software module that collects the results of prompting activities and records them as a log in a database.

[1197] "Generative AI model update means" refers to a means for updating the generative AI model based on the results of analysis and prompting activities, thereby improving the accuracy of future prompting plans.

[1198] This invention is a system for streamlining debt management and collection operations, collecting necessary information from multiple data sources, automating analysis and collection activities, and further performing emotional analysis, enabling flexible responses according to the emotional state of the debtor.

[1199] Data collection

[1200] The server collects data such as the debtor's payment history, communication history, and financial situation. Specifically, it obtains the necessary information using the credit bureau's API, the bank's transaction database, and the telecommunications company's API. The server sends requests to these data sources, for example, accessing the "credit bureau API" endpoint to obtain the debtor's payment history. Similarly, it uses an SQL query to obtain transaction history for the past year from the "bank transaction database," and obtains phone and email records via the telecommunications company's API. This data is then stored in a "data collection folder" by the server.

[1201] Data analysis and preprocessing

[1202] The server preprocesses the collected data, removing noise and filling in missing values. It uses a "data cleaning function" to fill in missing values ​​and applies a "noise removal algorithm" to refine the data. After this process is complete, the preprocessed data is input into a "machine learning model" to predict payment trends and delay risk. Specifically, it uses a "payment pattern identification model" to analyze past payment data and calculate the risk of the next delay.

[1203] Generate a customized reminder plan

[1204] Based on the analysis results, the server generates the optimal reminder plan for each debtor. To generate the plan, a "plan generation algorithm" is used to determine the optimal reminder method (telephone, email, written) and timing. For example, in the case of "Debtor A," if past data shows that the debtor has a high response rate to email reminders, email is selected as the top priority method and entered into the reminder plan table.

[1205] Utilizing the Emotion Engine

[1206] The server uses an "emotion engine" to recognize the debtor's emotions during collection activities. This engine detects emotions by applying a "voice analysis module" to phone recordings and a "text analysis module" to email and chat content. Emotional states (positive, negative, neutral) are recorded in an "emotion log" to help adjust collection plans.

[1207] Execute automatic reminders

[1208] The server automatically executes reminder activities based on the generated reminder plan. The "reminder email generation module" generates the email content and sends it at the specified date and time using the "email sending script." It also makes reminder calls via the "automatic call system" and plays back recorded messages. The reminder plan is adjusted in real time using emotional information recognized by the emotion engine.

[1209] Tracking lost contacts

[1210] The server tracks the latest contact information of unreachable debtors using the resident registration database and the API of the telecommunications company. Specifically, it sends a request to the "resident registration API" to obtain the latest address and contact information. This information is then updated in the "contact database."

[1211] Legal process assistance

[1212] The server automatically generates documents to support the necessary legal procedures. Mediation and bankruptcy documents are created based on the "document generation template" and formatted by the "PDF generation module." The documents are then ready to be sent to the relevant departments and lawyers.

[1213] Feedback and Optimization

[1214] The server collects the results of the reminder activities and records them in a "feedback database." For example, it stores logs of the open rate of emails sent, response status of phone calls, and whether or not payments were made. It also collects emotional information recognized by the emotion engine and updates the "generative AI model." This feedback loop improves the accuracy of future reminder plans.

[1215] Examples:

[1216] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[1217] Example prompt sentence:

[1218] "Calculate the optimal method and timing for reminding Debtor A based on their payment history and communication history over the past year. Then, optimize it using an emotion engine."

[1219] The above is a specific embodiment for carrying out the invention. This system automates the debt collection process and enables efficient and fast debt collection activities that take into account the emotional state of the user.

[1220] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1221] Step 1:

[1222] The server collects data such as the debtor's payment history, communication history, and financial situation. The API of the credit information agency, the bank's transaction database, and the API endpoint of the telecommunications company are used as input. The server then sends requests to each API, obtains the required data, and saves it in a data collection folder. Specifically, it sends an API request, converts the returned data into a specified format, and saves it.

[1223] Step 2:

[1224] The server preprocesses the collected data. The input is the various collected data, and based on that, it performs noise removal and missing value imputation. It uses data cleaning functions to properly impute missing values ​​and applies noise removal algorithms. The output of this step is preprocessed data, which is sent to the next analysis step. Specific operations include filtering outliers and imputing missing values ​​using statistical methods.

[1225] Step 3:

[1226] The server analyzes the preprocessed data. The input is the preprocessed data, and the payment trends and late payment risk are analyzed based on a machine learning model. The received data is analyzed using a payment pattern identification model to calculate the late payment risk for each debtor. The results are output as an analysis report and used in the next step. Specifically, past payment data is input into the model to calculate the predicted late payment risk.

[1227] Step 4:

[1228] The server generates a customized reminder plan based on the analysis results. The input is the analysis report, and the plan generation algorithm is used to determine the individual reminder methods and timing. The generated plan is saved in the reminder plan table and sent to the execution step. Specifically, the server sets the optimal approach method and timing for each debtor based on the analysis results obtained from the database.

[1229] Step 5:

[1230] The server automatically executes reminder activities based on the generated reminder schedule. The input is data from the reminder schedule table, and the reminder email generation module generates the email content and sends it at the date and time specified by the email sending script. It also executes telephone reminders via an automatic calling system and plays back recorded messages. Specifically, it automatically executes the email script and sends an emergency message via the telephone system.

[1231] Step 6:

[1232] The server tracks the latest contact information for unreachable debtors. The input is the debtor information to be tracked, and the latest address and contact information is obtained using the resident registration database and the API of the telecommunications company. The server then updates the contact information in the contact database with the new information. Specifically, it sends a request to the resident registration API and updates the database with the obtained information.

[1233] Step 7:

[1234] The server automatically generates documents to support the necessary legal procedures. The input is the data required for the legal procedures, and mediation procedure documents and bankruptcy procedure documents are created using document generation templates and formatted in the PDF generation module. The generated documents are sent to relevant departments and lawyers. Specifically, the required information is entered into the document template, which is then automatically generated in PDF format.

[1235] Step 8:

[1236] The server collects the results of the prompting activities and updates the model. The inputs are the results of the prompting activities (email open rate, phone response status, etc.) and emotion recognition results, and the logs are saved in a feedback database. Based on this, the generative AI model is updated to improve the accuracy of future prompting plans. Specifically, the analysis results are added to the model learning data, and the model is retrained to improve accuracy.

[1237] The above are the specific processing steps of this system.

[1238] (Application example 2)

[1239] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the headset type terminal 314 will be referred to as a "terminal."

[1240] In today's world, effective debt collection activities through communication with debtors are important, but traditional methods fail to fully consider the individual circumstances and emotions of debtors. Furthermore, the advertising industry faces challenges in accurately capturing users' interests and purchasing intent and delivering advertisements at the optimal time. To solve these challenges, a system is needed that analyzes the behavior and emotions of debtors and users and automatically responds optimally.

[1241] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1242] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for analyzing the collected data to generate an optimal collection schedule for each debtor; means for automatically carrying out collection activities for debtors based on the generated collection schedule; means for searching for and tracking the latest contact information for debtors who have become unreachable; means for generating documents to support necessary legal procedures; means for collecting the results of the collection activities and updating a model based on the analysis results; means for collecting user purchase history, browsing history, and communication history; means for preprocessing the collected data and analyzing user interests and purchasing intentions using a machine learning algorithm; means for generating an optimal advertising plan for each user based on the analysis results; means for analyzing user emotions using an emotion analysis engine and adjusting advertisements; means for automatically delivering advertisements to users based on the generated advertising plan; and means for collecting the results of advertisement delivery and optimizing subsequent advertising plans. This automates the debt collection process and advertisement delivery process, enabling more efficient and effective responses while taking into account the user's emotional state and behavioral patterns.

[1243] A "debtor" is a person who has borrowed money from a financial institution or money lender and is obligated to repay the loan.

[1244] "Payment history" refers to a record of the payments a debtor has made to date, including the date, time, amount, and method of each payment.

[1245] "Communication history" refers to a record of contacts and conversations made through the means of communication used by the debtor (such as telephone or email).

[1246] "Financial situation" refers to the debtor's financial condition, including income, expenses, assets, and liabilities.

[1247] "Data collection methods" refer to the methods and techniques used to compile the required information, including APIs and databases.

[1248] "Data analysis tools" refer to techniques and methods used to pre-process and subsequently analyze collected data, including, for example, noise reduction and machine learning algorithms.

[1249] A "repayment plan" refers to specific methods and schedules for encouraging debtors to repay.

[1250] "Automatic collection activity means" refers to a technique or method for automatically sending collections to debtors based on the generated collection plan.

[1251] "Contact information retrieval means" refers to a technique or method for identifying the most current contact information for a debtor who has become unreachable.

[1252] "Document generation means for supporting legal procedures" refers to techniques and methods for automatically creating documents to be used in necessary legal procedures.

[1253] "Model update methods" refer to techniques and methods for updating generative algorithms and machine learning models based on analysis results, improving accuracy from the next time onwards.

[1254] "Purchase history" refers to a list or record of products a user has purchased in the past.

[1255] "Browsing history" refers to a record of the web pages and content a user has viewed on the Internet.

[1256] A "machine learning algorithm" is a mathematical model or computational procedure used to make predictions or classifications based on collected data.

[1257] An "advertising plan" refers to a specific method and schedule for delivering optimal advertisements to users.

[1258] An "emotion analysis engine" is software or algorithms for detecting and analyzing a user's emotional state.

[1259] "Automatic advertisement distribution means" refers to a technique or method for automatically distributing advertisements to users in accordance with a generated advertising plan.

[1260] "Advertising delivery result optimization means" refers to techniques and methods for optimizing future advertising plans based on the results of advertising delivery.

[1261] System Overview

[1262] The system of the present invention automates the collection and analysis of data such as debtor payment history, communication history, and financial situation, and has the function of generating an optimal collection plan and automatically carrying out collection activities.It can also be applied to the advertising field, and has the function of delivering optimal advertisements based on users' purchase history, browsing history, and emotional information.The system consists of the following elements: server, terminal, and user.

[1263] Program Details

[1264] Hardware and software used

[1265] Hardware: Servers, smartphones, smart glasses, head-mounted displays

[1266] Software: Emotion engine (Emotion API), machine learning model (TensorFlow), database (MySQL), ad delivery management software (Ad Manager API)

[1267] Data collection

[1268] The server uses APIs and databases to collect data on the debtor's payment history, communication history, and financial status. For example, it obtains payment history for the past year from a bank database and call history from a telecommunications company's API.

[1269] Data analysis

[1270] The server preprocesses the collected data, removing noise and filling in missing values, and then analyzes the data using a machine learning algorithm (using TensorFlow) to predict the debtor's payment trends and the optimal timing for collection.

[1271] Generate reminder schedule

[1272] Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, for debtors for whom email reminders are effective, email reminders are sent first, and if there is no response, the server switches to phone reminders.

[1273] Utilizing the Emotion Engine

[1274] The server uses the emotion engine to analyze the debtor's emotions during collection activities, for example, by analyzing the debtor's tone of voice during a collection call, and if negative emotions are detected, adjust the collection plan accordingly.

[1275] Execute automatic reminders

[1276] Based on the generated reminder plan, the server automatically executes reminder activities, automatically generates reminder emails and sends them at the specified date and time, and also makes reminder calls via an automated calling system and plays recorded messages.

[1277] Collection of user purchasing and browsing history

[1278] The server collects the user's purchase history and browsing history using various APIs, such as the History API of a web browser or the API of a communication app.

[1279] Advertising plan generation and ad delivery

[1280] The server preprocesses the collected user data and uses machine learning algorithms to analyze the user's interests and purchasing intent. Based on the analysis results, it generates an optimal advertising plan for each user. Based on the generated advertising plan, the server automatically delivers advertisements to the user.

[1281] Result feedback and model optimization

[1282] The server records the results of ad delivery in a database and optimizes future ad plans. Ad click rates and purchase rates are fed back into the model to improve the ad delivery algorithm.

[1283] Specific examples

[1284] The server retrieves debtor A's payment history for the past year from the bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th of each month, and if there is no response, a reminder phone call is made. Similarly, the server analyzes user B's past purchase and browsing history and delivers the optimal advertisement during the lunch break.

[1285] Prompt Sentence Examples

[1286] "We analyze a user's purchasing and browsing history over the past year to predict what products they are likely to purchase next. We use the Emotion API to analyze their emotions while watching an ad, and then display ads from categories that they responded positively to."

[1287] This will automate the debt collection and advertising delivery processes, enabling more efficient and effective operations.

[1288] The flow of the specific processing in the application example 2 will be described with reference to FIG.

[1289] Step 1:

[1290] Data collection

[1291] The server uses APIs and databases to collect the necessary data.

[1292] Input: Bank database API, Telecommunications company API, Web browser History API, Telecommunications app API

[1293] Specific operation: Obtain the debtor's payment history for the past year from the bank database, obtain call history from the telecommunications company's API, and obtain the user's purchase history and browsing history using the web browser's History API and the communication app's API.

[1294] Output: Data set of payment history, communication history, purchase history, browsing history, etc. for the relevant debtors and users

[1295] Step 2:

[1296] Data Preprocessing

[1297] The server preprocesses the collected data, removing noise and completing missing values.

[1298] Input: The dataset collected in step 1

[1299] Specific actions: Identify and remove or correct inaccurate information in collected data, and impute missing values ​​in an appropriate manner.

[1300] Output: A clean and reliable dataset

[1301] Step 3:

[1302] Data analysis

[1303] The server analyzes the pre-processed data using machine learning algorithms to predict the debtor's payment trends and the optimal timing for reminders.

[1304] Input: The dataset preprocessed in step 2, the machine learning algorithm (TensorFlow)

[1305] How it works: It uses machine learning algorithms to analyze each debtor's payment habits and risk of late payments, and also predicts the next likely purchase based on the user's purchase and browsing history.

[1306] Output: Analysis results on payment trends, optimal reminder timing, user interests, and purchasing intent

[1307] Step 4:

[1308] Generate promotional and advertising plans

[1309] Based on the analysis results, the server generates an optimal reminder plan for each debtor and an optimal advertising plan for each user.

[1310] Input: Analysis results obtained in Step 3

[1311] Specific operation: Based on the analysis results, the system determines the optimal reminder method (telephone, email, written) and timing for each debtor and generates a reminder plan. It also determines the optimal advertising content and distribution timing for each user and generates an advertising plan.

[1312] Output: Reminder plan for each debtor, advertising plan for each user

[1313] Step 5:

[1314] Emotion analysis

[1315] The server uses an emotion engine to analyze the user's emotions during the promotional activities and advertisement viewing.

[1316] Input: Voice data during promotional activities and ad viewing, text data, emotion engine (Emotion API)

[1317] Specific operation: The emotion engine analyzes voice and text data to identify emotional states such as positive, negative, or neutral.

[1318] Output: Analysis results on user sentiment

[1319] Step 6:

[1320] Execution of automatic reminders and advertisement distribution

[1321] The server automatically executes promotional activities and advertisement distribution based on the generated plan.

[1322] Input: Promotion plan and advertising plan generated in step 4

[1323] Specific operations: Automatically generate and send reminder emails to debtors at the specified date and time. Also, make reminder phone calls via an automated calling system. Furthermore, distribute advertisements at the specified times based on the advertising plan.

[1324] Output: Reminders performed and ads delivered

[1325] Step 7:

[1326] Result feedback and optimization

[1327] The server collects the results of the promotional activities and advertisement distribution and optimizes the plan for the next time.

[1328] Input: Promotional activity result data, ad delivery result data

[1329] Specific operations: Results such as the open rate of reminder emails, response status of phone calls, click rates and purchase rates of ads are recorded in a database, and the model is retrained and optimized.

[1330] Output: Updated machine learning models and optimization data for next promotional and advertising plans

[1331] Through these steps, the system of the invention automates the debt collection and advertising distribution processes, achieving effective and efficient operations.

[1332] The specific processing unit 290 transmits the result of the specific processing to the headset type terminal 314. In the headset type terminal 314, the control unit 46A causes the speaker 240 and the display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating a user input regarding the result of the specific processing. The control unit 46A transmits audio data indicating the user input acquired by the microphone 238 to the data processing device 12. In the data processing device 12, the specific processing unit 290 acquires the audio data.

[1333] The data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of the data generation model 58 is ChatGPT (Internet Search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search <url: https: gemini.google.com ?hl="ja">) and other generation AIs. The data generation model 58 is obtained by performing deep learning on a neural network. A prompt including an instruction is input to the data generation model 58, and inference data such as voice data indicating voice, text data indicating text, and image data indicating an image is also input. The data generation model 58 performs inference on the input inference data in accordance with the instruction indicated by the prompt, and outputs the inference result in a data format such as voice data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1334] In the above embodiment, an example was given in which the specific processing is performed by the data processing device 12, but the technology of the present disclosure is not limited to this, and the specific processing may be performed by the headset type terminal 314.

[1335] [Fourth embodiment]

[1336] FIG. 7 shows an example of the configuration of a data processing system 410 according to the fourth embodiment.

[1337] 7, a data processing system 410 includes a data processing device 12 and a robot 414. An example of the data processing device 12 is a server.

[1338] The data processing device 12 includes a computer 22, a database 24, and a communication I / F 26. The computer 22 is an example of a "computer" according to the technology of the present disclosure. The computer 22 includes a processor 28, a RAM 30, and a storage 32. The processor 28, the RAM 30, and the storage 32 are connected to a bus 34. The database 24 and the communication I / F 26 are also connected to the bus 34. The communication I / F 26 is connected to a network 54. Examples of the network 54 include a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1339] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication I / F 44, and a control target 443. The computer 36 includes a processor 46, a RAM 48, and a storage 50. The processor 46, the RAM 48, and the storage 50 are connected to a bus 52. The microphone 238, the speaker 240, the camera 42, and the control target 443 are also connected to the bus 52.

[1340] The microphone 238 receives instructions and the like from the user 20 by receiving voice uttered by the user 20. The microphone 238 captures the voice uttered by the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio in accordance with instructions from the processor 46.

[1341] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an imaging element such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the surroundings of user 20 (for example, an imaging range defined by an angle of view equivalent to the field of vision of a typical healthy person).

[1342] The communication I / F 44 is connected to a network 54. The communication I / Fs 44 and 26 are responsible for the exchange of various information between the processor 46 and the processor 28 via the network 54. The exchange of various information between the processor 46 and the processor 28 using the communication I / Fs 44 and 26 is carried out in a secure state.

[1343] The control object 443 includes a display device, LEDs in the eyes, and motors for driving the arms, hands, and feet. The posture and gestures of the robot 414 are controlled by controlling the motors of the arms, hands, and feet. Some of the emotions of the robot 414 can be expressed by controlling these motors. In addition, the facial expressions of the robot 414 can also be expressed by controlling the light emission state of the LEDs in the eyes of the robot 414.

[1344] Fig. 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Fig. 8, in the data processing device 12, a specific process is performed by the processor 28. A specific process program 56 is stored in the storage 32.

[1345] The specific processing program 56 is an example of a "program" according to the technology of the present disclosure. The processor 28 reads the specific processing program 56 from the storage 32 and executes the read specific processing program 56 on the RAM 30. The specific processing is realized by the processor 28 operating as a specific processing unit 290 in accordance with the specific processing program 56 executed on the RAM 30.

[1346] The storage 32 stores a data generation model 58 and an emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1347] In the robot 414, the processor 46 performs the reception output process. A reception output program 60 is stored in the storage 50. The processor 46 reads the reception output program 60 from the storage 50 and executes the read reception output program 60 on the RAM 48. The reception output process is realized by the processor 46 operating as the control unit 46A in accordance with the reception output program 60 executed on the RAM 48.

[1348] Next, a description will be given of the specific processing performed by the specific processing unit 290 of the data processing device 12. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1349] 1. Data Collection

[1350] Subject: Server

[1351] The server first collects data such as the debtor's payment history, communication history, and financial situation. At this stage, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the email communication history is obtained from the email server.

[1352] 2. Data Analysis

[1353] Subject: Server

[1354] The server analyzes the collected data. When analyzing the data, it first preprocesses it to remove noise and fill in missing values. It then uses a machine learning algorithm to predict payment trends and the optimal timing for reminders. Specifically, it inputs past delay patterns into a model to predict the likelihood of the next delay.

[1355] 3. Generate a customized reminder plan

[1356] Subject: Server

[1357] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will create a plan to first remind that debtor by email.

[1358] 4. Execute automatic reminders

[1359] Subject: Server

[1360] The server automatically executes reminder activities based on the generated reminder plan. For example, it automatically generates and sends reminder emails at a specified date and time. If a reminder call is required, it makes a call via an automated calling system and plays a recorded message.

[1361] 5. Tracking Lost Contacts

[1362] Subject: Server

[1363] The server tracks the latest contact information of debtors who have become unreachable, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, allowing lost contact to be restored.

[1364] 6. Legal assistance

[1365] Subject: Server

[1366] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[1367] 7. Feedback and optimization

[1368] Subject: Server

[1369] The server collects the results of collection activities and updates the model. For example, it records the open rate of emails and the response status of phone calls, and reflects this in the next collection plan. This feedback loop improves the accuracy of future collection plans, enabling more efficient debt collection.

[1370] Examples:

[1371] For example, the server retrieves debtor A's payment history for the past year via the bank database and API, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a further reminder phone call is made. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[1372] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and enables more efficient and faster debt collection activities.

[1373] The processing flow will be explained below.

[1374] Step 1:

[1375] Subject: Server

[1376] The server collects data such as the debtor's payment history, communication history, and financial situation. This collection process obtains the necessary information using the API of a credit information agency, a bank's transaction database, and a telecommunications company's API. For example, the server sends a query to a bank database to obtain a specific debtor's payment history for the past year. It also obtains communication history (phone and email records) from a telecommunications company's API.

[1377] Step 2:

[1378] Subject: Server

[1379] The server preprocesses the collected data, including cleaning the data (removing noise) and imputing missing values. Specific operations include deleting incomplete data rows and imputing missing values ​​with the mean or median. It also converts data in non-standard formats into a standard format.

[1380] Step 3:

[1381] Subject: Server

[1382] The server analyzes the preprocessed data and extracts payment trends for each debtor. This analysis uses machine learning models to detect debtor delay patterns from past payment history. For example, the server can identify debtors who have had many late payments in the past and calculate the risk of their next payment delay.

[1383] Step 4:

[1384] Subject: Server

[1385] The server generates an optimal reminder plan for each debtor. The generated reminder plan includes the optimal reminder method (phone, email, written) and optimal timing. For example, if a debtor has responded well to emails in the past, the server generates a plan that prioritizes email reminders.

[1386] Step 5:

[1387] Subject: Server

[1388] The server automatically executes reminder activities based on the generated reminder plan. Specifically, it automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder calls via an automated calling system and plays recorded messages.

[1389] Step 6:

[1390] Subject: Server

[1391] The server tracks the latest contact information of debtors who have become unreachable. To do this, it uses a resident registration database or a telecommunications company's API to identify the latest address and contact information. For example, the server accesses a resident registration database to obtain the debtor's latest address information.

[1392] Step 7:

[1393] Subject: Server

[1394] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for delivery to relevant departments and lawyers.

[1395] Step 8:

[1396] Subject: Server

[1397] The server collects the results of the collection activities and records them in a database, for example, by saving information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made.

[1398] Step 9:

[1399] Subject: Server

[1400] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and incorporate new patterns and trends.

[1401] Example 1

[1402] Next, a description will be given of Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1403] In conventional debt collection systems, collection plans for each debtor are general and uniform, and collections are not optimally tailored to the circumstances of each debtor. This reduces collection efficiency and makes it difficult to track debtors when they cannot be contacted. Furthermore, preparing documents required for legal procedures is a manual process that consumes time and resources.

[1404] The specific processing by the specific processing unit 290 of the data processing device 12 in the first embodiment is realized by the following means.

[1405] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for preprocessing the collected data, removing noise, and completing missing values; means for analyzing the data using a machine learning algorithm and predicting the optimal timing and means of collection for each debtor; means for generating a collection plan customized for each debtor based on the prediction; means for automatically carrying out collection activities for debtors based on the generated collection plan; means for searching for and tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of collection activities and updating the model based on the analysis results. This automates the debt collection process, enables optimal collection activities tailored to the circumstances of individual debtors, efficient tracking when debtors cannot be contacted, and enables the rapid and accurate preparation of legal procedure documents.

[1406] A "debtor" is an individual or legal entity that has a financial obligation in the form of a debt, loan, or other form of debt.

[1407] "Payment history" is a record of past payments made by a debtor, including details such as due dates, amounts, and payment methods.

[1408] "Communication history" refers to a record of communication activities such as phone calls, emails, and messages between you and the debtor.

[1409] "Financial status" refers to information that indicates the debtor's current financial condition, and includes data on income, expenses, assets, liabilities, and the like.

[1410] "Data collection means" refers to a method or device by which the server obtains information such as the debtor's payment history, communication history, and financial situation.

[1411] "Data preprocessing" is the process of preparing collected data before analysis, such as removing noise from the data and filling in missing values.

[1412] "Machine learning algorithms" are statistical methods and models for finding patterns and knowledge in data.

[1413] An "analysis means" is a method or device for predicting a debtor's behavioral tendencies and the optimal timing for making a collection request by utilizing collected and preprocessed data.

[1414] A "demand plan" is a plan that determines the optimal means and timing of demands to prevent a debtor from delaying repayments.

[1415] A "customized reminder plan" is a reminder plan that is individually tailored based on each debtor's characteristics and behavior.

[1416] The "automatic prompting activity execution means" is a method or device that allows the server to automatically execute prompting means such as telephone calls or e-mails based on the generated prompting plan.

[1417] A "contact information tracking means" is a method or device for searching for and tracking the latest contact information of a debtor who has become unreachable.

[1418] "Legal procedure support means" refers to a method or device for generating documents related to necessary legal procedures and providing them to relevant departments and experts.

[1419] The "feedback collection means" is a method or device for collecting the results of prompting activities and reflecting them in the next prompting plan.

[1420] A "model updater" is a method or device for retraining an analytical model based on collected feedback data to improve its accuracy.

[1421] The present invention provides a system for automating and efficiently carrying out debt collection activities against specific debtors. A specific embodiment of this system will now be described.

[1422] Hardware and software used

[1423] This system operates in a server-based network environment. The specific hardware and software that supports each process in this system are listed below.

[1424] Server: Collects data, performs analytics, performs remediation activities, tracks contact information, assists with legal proceedings, gathers feedback, and updates the model.

[1425] Example of hardware used: A server equipped with a high-performance CPU and large-capacity RAM

[1426] Examples of software used: MongoDB (database), Pandas (data processing), Scikit-learn (machine learning), Twilio (automated calls), SMTP server (email sending)

[1427] Terminal: A device that interacts with a server and displays or inputs information.

[1428] Examples of hardware used: personal computers, tablets, smartphones

[1429] Specific processing of the system

[1430] Data collection

[1431] The server uses APIs from credit information agencies, banks, and telecommunications companies to collect data such as debtor payment history, communication history, and financial status. For example, the server obtains debtor A's payment history for the past year from the bank API and stores that data in MongoDB.

[1432] Data Preprocessing

[1433] The collected data is first preprocessed. The server uses the Pandas library to remove noise and impute missing values, then formats the collected data into a data frame.

[1434] Data analysis

[1435] Machine learning algorithms are applied to the preprocessed data. The server uses Scikit-learn and models such as random forests to predict the debtor's payment trends and the optimal timing for reminders. For example, it analyzes the debtor's delay patterns and predicts the likelihood of the next delay.

[1436] Generate a customized reminder plan

[1437] Based on the analysis results, the server generates a customized reminder plan for each debtor. This reminder plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data shows that a debtor is likely to respond to email reminders, the server will plan to first remind that debtor by email.

[1438] Execute automatic reminders

[1439] Based on the generated reminder plan, the server automatically executes reminder activities. It automatically generates reminder emails at the specified date and time and sends them via the SMTP server. In addition, if a phone reminder is required, it makes an automated call via Twilio's API and plays a pre-recorded message.

[1440] Tracking lost contacts

[1441] For debtors who can no longer be contacted, the server uses the resident registration database or the telecommunications company's API to search for the latest contact information and update the database.

[1442] Legal process assistance

[1443] The server automatically generates documents to support necessary legal procedures, such as arbitration documents and bankruptcy documents, and sends them to the relevant departments and lawyers.

[1444] Gathering feedback and updating the model

[1445] The server records the results of each reminder activity (e.g., email open rates and phone call response rates) and collects them as feedback data. The collected feedback data is used to retrain the analysis model and improve the accuracy of the next reminder plan.

[1446] Specific examples

[1447] For example, the server retrieves debtor A's payment history for the past year from the bank API and stores the data in JSON format in MongoDB. Next, it uses the Pandas library to fill in missing values ​​in the data and uses Scikit-learn to predict payment trends using a random forest model. Based on the analysis results, it determines the optimal method of reminding debtor A, automatically generates and sends a reminder email on the 15th. If there is no response, it makes a reminder phone call using Twilio's API. The results are recorded in the database and reflected in the next model update.

[1448] Prompt Sentence Examples

[1449] "Based on Debtor A's payment history, email communication history, and phone call history for the past year, please predict the risk of delay in the next debt payment and plan the optimal method and timing of reminder payments."

[1450] The flow of the identification process in the first embodiment will be described with reference to FIG.

[1451] Program processing steps

[1452] Step 1: Data collection

[1453] Subject: Server

[1454] The server receives the debtor's identification information as input. The server collects data such as the debtor's payment history, communication history, and financial status through the APIs of credit bureaus, banks, and telecommunications companies. For example, the server obtains the payment history of debtor A through an API call and saves the data in JSON format. During this process, the server stores the obtained data in a database.

[1455] Input: Debtor Identification Information

[1456] Output: Data on debtor's payment history, communication history, financial status, etc. (JSON format)

[1457] Step 2: Data Preprocessing

[1458] Subject: Server

[1459] The server receives the collected data as input. It uses the Pandas library to preprocess the data, removing noise, imputing missing values, and preparing it in a format suitable for analysis. During this process, it generates a data frame and performs operations such as removing outliers.

[1460] Input: Collected data (JSON format)

[1461] Output: Preprocessed data (data frame)

[1462] Step 3: Data analysis

[1463] Subject: Server

[1464] The server receives preprocessed data as input. It uses Scikit-learn to apply machine learning algorithms (e.g., random forests) to predict payment trends and optimal reminder timing. The analysis results are output as predictions for each debtor, including the likelihood of the next late payment and the optimal reminder method.

[1465] Input: Preprocessed data (data frame)

[1466] Output: Analysis results (payment trends, optimal reminder timing)

[1467] Step 4: Generate a customized reminder plan

[1468] Subject: Server

[1469] The server receives the analysis results as input. Based on this, it generates a customized reminder plan for each debtor. For example, if the analysis shows that email reminders are effective for debtor A, the server determines the content and timing of the emails.

[1470] Input: Analysis results (payment trends, optimal reminder timing)

[1471] Output: Customized reminder plan

[1472] Step 5: Run automatic reminders

[1473] Subject: Server

[1474] The server takes your customized reminder plan as input and automatically executes reminder activities at the specified dates and times, for example, generating and sending reminder emails using an SMTP server, and optionally placing automated calls using Twilio's APIs to play pre-recorded messages.

[1475] Input: Customized reminder plan

[1476] Output: Reminder actions taken (emails sent, call logs)

[1477] Step 6: Track down the missing person

[1478] Subject: Server

[1479] The server receives contact information for unreachable debtors as input, and uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information and update the database.

[1480] Input: Unreachable person's contact information

[1481] Output: Latest updated contact information

[1482] Step 7: Legal assistance

[1483] Subject: Server

[1484] The server receives the necessary legal procedure information as input, and then automatically generates mediation and bankruptcy procedure documents based on templates and sends them to the relevant departments and lawyers.

[1485] Input: Legal process information and templates

[1486] Output: Auto-generated legal document

[1487] Step 8: Gather feedback and update the model

[1488] Subject: Server

[1489] The server receives as input the results of each reminder activity (e.g., email open rates and phone response rates), which are collected as feedback data and used to retrain the analytical model, improving the accuracy of the next reminder plan.

[1490] Input: Results of reminder activities (open rate, response status)

[1491] Output: Updated analytical model

[1492] These are the specific processing steps of this system's program. Through this process, collection activities can be optimized for each debtor, greatly improving the efficiency of debt collection.

[1493] (Application example 1)

[1494] Next, a description will be given of Application Example 1. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1495] In the debt collection process, debtors frequently experience payment delays and are unable to be contacted, resulting in a decline in collection efficiency. Other issues include the time and cost required to simultaneously contact multiple debtors and the preparation of legal paperwork. Furthermore, it can be difficult to determine the optimal timing for collection, which can result in the ineffectiveness of payment collection efforts. These issues need to be resolved.

[1496] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 1 is realized by the following means.

[1497] In this invention, the server includes means for collecting data such as debtor payment history, communication history, and financial status, means for analyzing the collected data to generate an optimal collection plan for each debtor, means for automatically carrying out collection activities for debtors based on the generated collection plan, means for searching for and tracking the latest contact information for debtors who have become unreachable, means for generating documents to support necessary legal procedures, means for collecting the results of the collection activities and updating the model based on the analysis results, means for predicting payment trends and determining the optimal timing for collection activities to prevent payment delays, and means for automatically generating and sending notification messages at specified dates and times. This streamlines the debt collection process and enables fast and effective debt collection activities.

[1498] A "debtor" is an individual or corporation that is obligated to borrow money from a financial institution, business partner, etc., or to pay for goods or services.

[1499] A "payment history" is a record of past payments made by a debtor, including details such as the date, amount, and method of payment within a specific period of time.

[1500] "Communication history" refers to a record of the debtor's communication activities, such as phone calls, emails, and messages, including the date, time, frequency, and content of the messages sent and received.

[1501] "Financial situation" refers to information that shows the debtor's economic status, such as income, expenses, assets, and liabilities.

[1502] "Means of collecting data" refers to the mechanisms and technologies for collecting the necessary data using APIs from credit information agencies, bank transaction databases, telecommunications company APIs, etc.

[1503] "Means for analyzing data" refers to methods and techniques for preprocessing collected data and using machine learning models to predict payment trends and find the optimal timing for reminders.

[1504] A "demand collection plan" is a plan that specifically outlines when and how to request payment from each debtor.

[1505] "Means for carrying out collection activities" refers to systems or technologies that automatically notify or contact debtors based on the generated collection plan.

[1506] "Means for searching and tracking contact information" refers to methods and technologies for identifying the latest address and contact information of a debtor who has lost contact, using resident registration databases or telecommunications company APIs.

[1507] "Means for generating documents to support legal procedures" refers to systems and technologies that automatically create documents necessary for legal responses such as arbitration procedures and bankruptcy procedures.

[1508] "Means for collecting the results of prompting activities and updating the model" refers to methods and techniques for evaluating the effectiveness of prompting activities, improving the analytical model based on that data, and increasing the accuracy of future prompting plans.

[1509] "Measures to prevent payment delays" refers to methods and technologies that use machine learning models to predict payment trends and take appropriate action in advance if there is a high possibility of a delay.

[1510] "Means for generating and sending notification messages" refers to systems or technologies for automatically creating and sending payment reminder messages to debtors at specified dates and times.

[1511] The following describes an embodiment of the present invention.

[1512] The server first collects data such as the debtor's payment history, communication history, and financial situation. In this step, the necessary information is obtained using the API of a credit information agency, a bank's transaction database, a telecommunications company's API, etc. For example, for a specific debtor, the payment history for the past year is obtained from a bank database, and the communication history is obtained from a telecommunications company's API.

[1513] The server then analyzes the collected data. During data analysis, the data is preprocessed to remove noise and fill in missing values. After that, a machine learning algorithm is used to predict payment trends and the optimal timing for reminders. Specifically, past delay patterns are input into the model to predict the likelihood of the next delay.

[1514] Based on the analysis results, the server generates a customized reminder plan for each debtor. This plan includes the optimal reminder method (SNS notification, app notification, email) and the optimal timing. For example, if past data shows that a particular debtor is likely to respond to SNS notifications, the server will create a plan to first remind that debtor via SNS notification.

[1515] Based on the generated reminder plan, the server automatically generates and sends notification messages at the specified date and time. For example, when a payment deadline approaches, a notification message is automatically sent to the debtor's smartphone. The server also records the notification opening rate and payment completion rate to evaluate the effectiveness of the notifications.

[1516] For debtors who have become unreachable, the server tracks their latest contact information, using resident registration databases and telecommunications company APIs to identify their latest addresses and contact details, allowing for lost contact to be reestablished and effective collection efforts to continue.

[1517] In addition, the server automatically generates documents required for mediation procedures, bankruptcy procedures, etc. to assist with necessary legal procedures, and the generated documents are sent to relevant departments and lawyers.

[1518] The server collects the results of debt collection activities and updates the model to reflect them in the next collection plan. For example, it analyzes email open rates and phone response status, and improves the accuracy of the model through a feedback loop.

[1519] Hardware and software used

[1520] In this system, software is installed on the server to collect data via the credit information agency's API, the bank's transaction database, and the telecommunications company's API. Python libraries such as pandas and scikit-learn are used for data analysis, and smtplib is used to generate and send notification messages. RandomForestClassifier is used as the machine learning model.

[1521] Specific examples

[1522] For example, the payment history of a certain debtor A over the past year is collected, and a machine learning model is used to predict the likelihood of the next payment being late. Based on the analysis results, an SNS notification is selected for Debtor A, and a notification message is sent the day before the payment due date. This notification message is automatically generated, and the open rate and response status after sending are recorded. This data is reflected in the next reminder plan.

[1523] Prompt Sentence Examples

[1524] "Please predict the likelihood of User A's next payment being late based on their bank payment history, communication history, and financial situation over the past year. Based on the results, please provide a Python script that will generate an optimal reminder plan and automate the process of sending a notification message to User A."

[1525] The flow of the specific processing in the application example 1 will be described with reference to FIG.

[1526] Step 1:

[1527] Data collection: The server collects data such as the debtor's payment history, communication history, and financial situation via the API of credit information agencies, bank transaction databases, telecommunications company APIs, etc. Specifically, for a specific debtor, the payment history for the past year is obtained from the bank database, and the communication history is obtained from the telecommunications company API.

[1528] Input: Debtor identification information (e.g., Debtor ID)

[1529] Output: Datasets such as payment history, communication history, and financial status

[1530] Step 2:

[1531] Data preprocessing: The server preprocesses the collected data, removing noise and imputing missing values. This preprocessing cleans the data for later analysis. For example, missing values ​​are imputed with the mean value and outliers are removed.

[1532] Input: Collected dataset

[1533] Output: A preprocessed and clean dataset

[1534] Step 3:

[1535] Data analysis: The server analyzes the preprocessed data. It uses the machine learning algorithm RandomForestClassifier to predict payment trends and the optimal timing for reminders. It also inputs past delay patterns into the model to predict the likelihood of the next delay.

[1536] Input: Preprocessed clean dataset

[1537] Output: Predictions regarding payment trends and likelihood of delays

[1538] Step 4:

[1539] Generation of reminder plan: Based on the analysis results, the server generates an optimal reminder plan for each debtor. For example, if it is determined that a particular debtor is likely to respond to SNS notifications, it will create a reminder plan for that debtor via SNS notifications.

[1540] Input: Prediction result

[1541] Output: Optimal reminder plan

[1542] Step 5:

[1543] Generation and sending of notification messages: The server automatically generates and sends notification messages at the specified date and time based on the generated reminder plan. For example, a reminder message can be sent to the debtor's smartphone the day before the payment deadline.

[1544] Input: Optimal reminder schedule

[1545] Output: Notification message sent

[1546] Step 6:

[1547] Contact information tracking: For debtors who have become unable to contact us, the server uses resident registration databases and telecommunications company APIs to identify their latest addresses and contact information, thereby restoring lost contact.

[1548] Input: Uncontactable debtor information

[1549] Output: Latest contact information

[1550] Step 7:

[1551] Generation of documents to support legal procedures: The server automatically generates documents required for necessary legal procedures (e.g., arbitration procedures, bankruptcy procedures) and sends them to the relevant departments and lawyers.

[1552] Input: Debtor's Legal Information

[1553] Output: Auto-generated legal document

[1554] Step 8:

[1555] Model update: The server collects the results of the reminder activities and updates the analytical model based on them. This improves the accuracy of the next reminder plan. For example, the server records the opening rate of notification messages and the payment completion rate and reflects them in the model.

[1556] Input: Result data of reminder activity

[1557] Output: Updated analytical model

[1558] Furthermore, an emotion engine that estimates the user's emotion may be further combined. That is, the identification processing unit 290 may estimate the user's emotion using the emotion identification model 59, and perform identification processing using the user's emotion.

[1559] 1. Data Collection

[1560] Subject: Server

[1561] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains the payment history for the past year from the bank database, and the communication history (phone and email records) from the telecommunications company's API.

[1562] 2. Data Analysis

[1563] Subject: Server

[1564] The server preprocesses the collected data, removing noise and filling in missing values. It then uses machine learning algorithms to predict payment trends and the optimal timing for reminders. The server inputs past delay patterns into the model and calculates the risk of the next delay.

[1565] 3. Generate a customized reminder plan

[1566] Subject: Server

[1567] Based on the analysis results, the server generates an optimal reminder plan for each debtor. This plan includes the optimal reminder method (phone, email, written) and the optimal timing. For example, if past data indicates that a debtor is likely to respond to email reminders, the server will first plan to remind that debtor by email.

[1568] 4. Utilizing the Emotion Engine

[1569] Subject: Server

[1570] The server uses an emotion engine to recognize the user's (debtor's) emotions during collection activities. The emotion engine detects emotions through voice and text analysis and identifies emotional states such as positive, negative, and neutral. For example, during a phone call, the server analyzes emotions from the user's tone of voice and the content of their text replies.

[1571] 5. Execute automatic reminders

[1572] Subject: Server

[1573] The server automatically executes reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. It also makes reminder calls via an automated calling system and plays recorded messages as needed. The emotion information recognized by the emotion engine is used to adjust the reminder plan.

[1574] 6. Tracking Lost Contacts

[1575] Subject: Server

[1576] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest address and contact information using a resident registration database or a telecommunications company's API. For example, it accesses a resident registration database to obtain the debtor's latest address information.

[1577] 7. Legal assistance

[1578] Subject: Server

[1579] The server automatically generates documents to support the necessary legal procedures, such as mediation and bankruptcy procedure documents, based on templates, and prepares them for sending to the relevant departments and lawyers.

[1580] 8. Feedback and optimization

[1581] Subject: Server

[1582] The server collects the results of collection activities and records them in a database. Information such as the open rate of emails sent, response status of phone calls, and whether or not payments were made is saved as a log. In addition, emotional information recognized by the emotion engine is also collected, and the generative AI model is updated based on the analysis results. This feedback loop improves the accuracy of collection plans for future collections, enabling more efficient debt collection.

[1583] Examples:

[1584] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[1585] The above is a specific embodiment for carrying out the invention, which automates the debt collection process and allows for more efficient and faster debt collection activities while taking into account the emotional state of the user.

[1586] The processing flow will be explained below.

[1587] Step 1:

[1588] Subject: Server

[1589] The server collects data such as the debtor's payment history, communication history, and financial situation. It obtains the necessary information using the credit information agency's API, the bank's transaction database, and the telecommunications company's API. For example, the server obtains Debtor A's payment history for the past year from the bank database, and collects Debtor A's phone and email history for the past six months through the telecommunications company's API.

[1590] Step 2:

[1591] Subject: Server

[1592] The server preprocesses the collected data, which includes cleaning the data (removing noise) and imputing missing values. For example, the server removes incomplete data rows, imputes missing values ​​with the mean or median, and converts non-standardized data into a standardized format.

[1593] Step 3:

[1594] Subject: Server

[1595] The server analyzes the preprocessed data and extracts payment trends for each borrower. This analysis uses machine learning models to detect patterns of late payments from borrowers based on their past payment history. For example, the server can identify borrowers who have frequently made late payments in the past and predict the risk of the next late payment.

[1596] Step 4:

[1597] Subject: Server

[1598] The server generates an optimal reminder plan for each debtor. The generated plan includes the optimal reminder method (telephone, email, written) and optimal timing. Specifically, if a debtor has responded well to email reminders in the past, the server generates a plan that prioritizes email reminders.

[1599] Step 5:

[1600] Subject: Server

[1601] The server uses an emotion engine to recognize the user's emotions during the prompting activity. The emotion engine detects emotions through voice analysis and text analysis. For example, during a phone prompting call, the server can determine the user's emotional state (positive, negative, neutral, etc.) from the tone and speed of the user's voice and the content of the message.

[1602] Step 6:

[1603] Subject: Server

[1604] The server adjusts the reminder plan in real time based on the user's emotional information recognized by the emotion engine. For example, if the user shows negative emotions, the server may soften the tone of the reminder or temporarily suspend the reminder.

[1605] Step 7:

[1606] Subject: Server

[1607] The server automatically carries out reminder activities based on the generated reminder plan. It automatically generates reminder emails and sends them at the specified date and time. If necessary, it also makes reminder phone calls via an automated calling system and plays recorded messages. For example, the server automatically generates and sends a reminder email to debtor A on the 15th, and if there is no response, it makes a reminder phone call.

[1608] Step 8:

[1609] Subject: Server

[1610] The server tracks the latest contact information of debtors who have become unreachable. It identifies their latest addresses and contact information using a resident registration database or a telecommunications company's API. For example, it accesses the resident registration database to obtain the latest address information of Debtor A.

[1611] Step 9:

[1612] Subject: Server

[1613] The server automatically generates documents to support the necessary legal procedures. Mediation procedure documents and bankruptcy procedure documents are automatically created based on templates and sent to the relevant departments and lawyers. For example, the server automatically generates mediation procedure documents for Debtor A and prepares them to be sent to the lawyer.

[1614] Step 10:

[1615] Subject: Server

[1616] The server collects the results of the reminder activities and records them in a database. It stores information such as the open rate of emails sent, the response status of phone calls, and whether or not payments were made as logs. It also collects emotional information recognized by the emotion engine.

[1617] Step 11:

[1618] Subject: Server

[1619] The server updates the generative AI model based on the collected data, improving the accuracy of future reminder plans. Specifically, it uses machine learning algorithms to retrain the model and apply new patterns and trends to it.

[1620] Example 2

[1621] Next, a description will be given of Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1622] Conventional debt collection systems have the problem of being inefficient and time-consuming, requiring a lot of manual work. It is also difficult to implement collection activities that take into account the debtor's feelings, and there is a lack of appropriate timing and means for approaching debtors. Furthermore, tracking down debtors who have become unreachable and generating the necessary documents for legal procedures are often done manually, which is labor-intensive and time-consuming. There is a need for a system that can solve these problems and achieve efficient and effective debt collection.

[1623] The specific processing by the specific processing unit 290 of the data processing device 12 in the second embodiment is realized by the following means.

[1624] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial situation; means for preprocessing the collected data to remove noise and fill in missing values; means for analyzing the preprocessed data and generating an optimal collection plan for each debtor; means for recognizing the debtor's emotions based on the analysis results and generating a customized collection plan; means for automatically conducting loan collection activities based on the generated collection plan; means for tracking the latest contact information of debtors who have become unreachable; means for generating documents to support necessary legal procedures; and means for collecting the results of the collection activities and updating the generation AI model based on the analysis results. This automates the debt collection process and enables efficient and prompt debt collection activities that take into account the user's emotional state.

[1625] "Data collection tools" refers to hardware and software used to obtain data such as debtors' payment history, communication history, and financial status.

[1626] The "preprocessing means" refers to a software module that has the function of removing noise and filling in missing values ​​from collected data.

[1627] "Data Analysis Tools" means machine learning algorithms and related software used to analyze pre-processed data and predict debtor behavior patterns and optimal collection strategies.

[1628] The "demand plan generation means" refers to software and algorithms that determine the means and timing of demands customized for each debtor based on the analysis results and generate a plan.

[1629] "Emotion Recognition Means" means software modules and algorithms for recognizing the emotional state of a debtor through speech and text analysis.

[1630] "Automated collection measures" refers to software and hardware that automates loan collection activities using email and automated calling systems based on generated collection plans.

[1631] "Tracking means" refers to software and hardware that has the ability to obtain the latest contact information of debtors who have become unreachable by using resident registration databases or telecommunications company APIs.

[1632] "Document Generation Measures" means software modules and templates for automatically generating arbitration documents and bankruptcy documents to support necessary legal proceedings.

[1633] The "feedback collection means" is a software module that collects the results of prompting activities and records them as a log in a database.

[1634] "Generative AI model update means" refers to a means for updating the generative AI model based on the results of analysis and prompting activities, thereby improving the accuracy of future prompting plans.

[1635] This invention is a system for streamlining debt management and collection operations, collecting necessary information from multiple data sources, automating analysis and collection activities, and further performing emotional analysis, enabling flexible responses according to the emotional state of the debtor.

[1636] Data collection

[1637] The server collects data such as the debtor's payment history, communication history, and financial situation. Specifically, it obtains the necessary information using the credit bureau's API, the bank's transaction database, and the telecommunications company's API. The server sends requests to these data sources, for example, accessing the "credit bureau API" endpoint to obtain the debtor's payment history. Similarly, it uses an SQL query to obtain transaction history for the past year from the "bank transaction database," and obtains phone and email records via the telecommunications company's API. This data is then stored in a "data collection folder" by the server.

[1638] Data analysis and preprocessing

[1639] The server preprocesses the collected data, removing noise and filling in missing values. It uses a "data cleaning function" to fill in missing values ​​and applies a "noise removal algorithm" to refine the data. After this process is complete, the preprocessed data is input into a "machine learning model" to predict payment trends and delay risk. Specifically, it uses a "payment pattern identification model" to analyze past payment data and calculate the risk of the next delay.

[1640] Generate a customized reminder plan

[1641] Based on the analysis results, the server generates the optimal reminder plan for each debtor. To generate the plan, a "plan generation algorithm" is used to determine the optimal reminder method (telephone, email, written) and timing. For example, in the case of "Debtor A," if past data shows that the debtor has a high response rate to email reminders, email is selected as the top priority method and entered into the reminder plan table.

[1642] Utilizing the Emotion Engine

[1643] The server uses an "emotion engine" to recognize the debtor's emotions during collection activities. This engine detects emotions by applying a "voice analysis module" to phone recordings and a "text analysis module" to email and chat content. Emotional states (positive, negative, neutral) are recorded in an "emotion log" to help adjust collection plans.

[1644] Execute automatic reminders

[1645] The server automatically executes reminder activities based on the generated reminder plan. The "reminder email generation module" generates the email content and sends it at the specified date and time using the "email sending script." It also makes reminder calls via the "automatic call system" and plays back recorded messages. The reminder plan is adjusted in real time using emotional information recognized by the emotion engine.

[1646] Tracking lost contacts

[1647] The server tracks the latest contact information of unreachable debtors using the resident registration database and the API of the telecommunications company. Specifically, it sends a request to the "resident registration API" to obtain the latest address and contact information. This information is then updated in the "contact database."

[1648] Legal process assistance

[1649] The server automatically generates documents to support the necessary legal procedures. Mediation and bankruptcy documents are created based on the "document generation template" and formatted by the "PDF generation module." The documents are then ready to be sent to the relevant departments and lawyers.

[1650] Feedback and Optimization

[1651] The server collects the results of the reminder activities and records them in a "feedback database." For example, it stores logs of the open rate of emails sent, response status of phone calls, and whether or not payments were made. It also collects emotional information recognized by the emotion engine and updates the "generative AI model." This feedback loop improves the accuracy of future reminder plans.

[1652] Examples:

[1653] For example, the server retrieves debtor A's payment history for the past year from a bank database, then analyzes the optimal means and timing of reminders to debtor A to generate a reminder plan. Based on the generated plan, a reminder email is automatically generated and sent to debtor A on the 15th, and if there is no response, a follow-up phone call is made. At this time, the emotion engine recognizes the emotion from the debtor's tone of voice and adjusts the reminder plan. The results of these reminder activities are then recorded in the database and reflected in the next reminder plan, optimizing the model.

[1654] Example prompt sentence:

[1655] "Calculate the optimal method and timing for reminding Debtor A based on their payment history and communication history over the past year. Then, optimize it using an emotion engine."

[1656] The above is a specific embodiment for carrying out the invention. This system automates the debt collection process and enables efficient and fast debt collection activities that take into account the emotional state of the user.

[1657] The flow of the identification process in the second embodiment will be described with reference to FIG.

[1658] Step 1:

[1659] The server collects data such as the debtor's payment history, communication history, and financial situation. The API of the credit information agency, the bank's transaction database, and the API endpoint of the telecommunications company are used as input. The server then sends requests to each API, obtains the required data, and saves it in a data collection folder. Specifically, it sends an API request, converts the returned data into a specified format, and saves it.

[1660] Step 2:

[1661] The server preprocesses the collected data. The input is the various collected data, and based on that, it performs noise removal and missing value imputation. It uses data cleaning functions to properly impute missing values ​​and applies noise removal algorithms. The output of this step is preprocessed data, which is sent to the next analysis step. Specific operations include filtering outliers and imputing missing values ​​using statistical methods.

[1662] Step 3:

[1663] The server analyzes the preprocessed data. The input is the preprocessed data, and the payment trends and late payment risk are analyzed based on a machine learning model. The received data is analyzed using a payment pattern identification model to calculate the late payment risk for each debtor. The results are output as an analysis report and used in the next step. Specifically, past payment data is input into the model to calculate the predicted late payment risk.

[1664] Step 4:

[1665] The server generates a customized reminder plan based on the analysis results. The input is the analysis report, and the plan generation algorithm is used to determine the individual reminder methods and timing. The generated plan is saved in the reminder plan table and sent to the execution step. Specifically, the server sets the optimal approach method and timing for each debtor based on the analysis results obtained from the database.

[1666] Step 5:

[1667] The server automatically executes reminder activities based on the generated reminder schedule. The input is data from the reminder schedule table, and the reminder email generation module generates the email content and sends it at the date and time specified by the email sending script. It also executes telephone reminders via an automatic calling system and plays back recorded messages. Specifically, it automatically executes the email script and sends an emergency message via the telephone system.

[1668] Step 6:

[1669] The server tracks the latest contact information for unreachable debtors. The input is the debtor information to be tracked, and the latest address and contact information is obtained using the resident registration database and the API of the telecommunications company. The server then updates the contact information in the contact database with the new information. Specifically, it sends a request to the resident registration API and updates the database with the obtained information.

[1670] Step 7:

[1671] The server automatically generates documents to support the necessary legal procedures. The input is the data required for the legal procedures, and mediation procedure documents and bankruptcy procedure documents are created using document generation templates and formatted in the PDF generation module. The generated documents are sent to relevant departments and lawyers. Specifically, the required information is entered into the document template, which is then automatically generated in PDF format.

[1672] Step 8:

[1673] The server collects the results of the prompting activities and updates the model. The inputs are the results of the prompting activities (email open rate, phone response status, etc.) and emotion recognition results, and the logs are saved in a feedback database. Based on this, the generative AI model is updated to improve the accuracy of future prompting plans. Specifically, the analysis results are added to the model learning data, and the model is retrained to improve accuracy.

[1674] The above are the specific processing steps of this system.

[1675] (Application example 2)

[1676] Next, a description will be given of Application Example 2. In the following description, the data processing device 12 will be referred to as a "server" and the robot 414 will be referred to as a "terminal."

[1677] In today's world, effective debt collection activities through communication with debtors are important, but traditional methods fail to fully consider the individual circumstances and emotions of debtors. Furthermore, the advertising industry faces challenges in accurately capturing users' interests and purchasing intent and delivering advertisements at the optimal time. To solve these challenges, a system is needed that analyzes the behavior and emotions of debtors and users and automatically responds optimally.

[1678] The specific processing by the specific processing unit 290 of the data processing device 12 in the application example 2 is realized by the following means.

[1679] In this invention, the server includes: means for collecting data such as debtor payment history, communication history, and financial status; means for analyzing the collected data to generate an optimal collection schedule for each debtor; means for automatically carrying out collection activities for debtors based on the generated collection schedule; means for searching for and tracking the latest contact information for debtors who have become unreachable; means for generating documents to support necessary legal procedures; means for collecting the results of the collection activities and updating a model based on the analysis results; means for collecting user pur...

Claims

1. A means of collecting data such as debtor's payment history, communication history, financial status, etc.; A means for analyzing the collected data and generating an optimal collection plan for each debtor; a means for automatically carrying out collection activities against debtors based on the generated collection plan; A means to find and track current contact information for debtors who have become unreachable; a means of generating documents to support necessary legal proceedings; A means of collecting the results of the prompting activities and updating the model based on the analysis results; A system including:

2. The system of claim 1 , wherein the system performs data cleaning and pre-processing.

3. 2. The system according to claim 1, wherein the system predicts the optimal timing for making a payment based on a specific behavioral pattern of the debtor.

Images