Method and apparatus for personal credit assessment based on non-financial bond data

Abstract

A server providing a personal credit rating service based on non-financial bond data can acquire source data for credit rating that includes information on the relationship between the creditor and the debtor, the purpose of the debt, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, nationality, whether the individual resides overseas, the number of overdue payments, and occupation and age information. The server can perform normalization and structuring processing on the source data, including natural language processing-based text normalization, rule-based parsing, outlier removal, and categorical encoding. The server can generate one or more derived variables representing repayment propensity, risk factors, lifestyle-based signals, and behavior-based signals from the normalized and structured data. The server can evaluate an individual's credit risk through an artificial intelligence agent that takes the derived variables and the normalized and structured data as input. The server can calculate an individual's credit score based on the evaluated credit risk.

Description

Technology Field

[0001] This specification relates to a technology for evaluating an individual's creditworthiness based on various debt and credit-related data generated in the non-financial sector. More specifically, it relates to an artificial intelligence-based credit evaluation method and apparatus for calculating or predicting an individual's creditworthiness by utilizing unstructured and non-financial data, such as information on the relationship between the creditor and the debtor, the level of debtor information held by the creditor, the purpose of the debt, whether the debtor is out of contact, whether the debtor is undergoing personal rehabilitation or bankruptcy, nationality, whether the debtor resides overseas, the number of overdue payments, the contract holder's occupation, supporting documents held by the creditor, and the source of funds. Background Technology

[0002] Existing personal credit evaluation methods have primarily relied on scoring systems derived from structured data handled within the financial sector, such as loan transaction history, repayment records, delinquency records, credit card usage, and financial asset holdings. While this approach can ensure a certain level of evaluation accuracy for users with sufficient experience in financial transactions, it may face limitations in precisely assessing creditworthiness for individuals with insufficient financial transaction history or those whose economic activities are centered in non-financial sectors. In particular, data arising from personal debt relationships, small-scale debt transactions, informal borrowing, and daily economic activities is often not reflected in financial institutions' credit evaluation systems; consequently, there is a possibility that evaluation results may differ from actual circumstances, even for individuals with the same level of risk.

[0003] Meanwhile, debt data or personal lifestyle information generated from non-financial sectors may contain behavioral signals distinct from financial transaction data, and this information has the potential to serve as meaningful supplementary data for assessing a debtor's creditworthiness. For instance, factors such as the relationship between the creditor and the debtor, the level of debt information held by the creditor, the purpose of the debt, whether the debtor is out of contact, history of personal rehabilitation or bankruptcy, overseas residency status, number of overdue payments, and the contract holder's occupation and age can provide significant insights into a debtor's sense of responsibility, willingness to repay, or risk factors, even though these are difficult for the financial sector to collect. Such data can exist in various forms, including not only structured information but also question-and-answer interviews, on-site verification results, unstructured text, and online community activities, and represents information that was difficult to utilize in traditional credit rating models.

[0004] Recently, advancements in artificial intelligence and unstructured data analysis technologies have drawn attention to the potential of personal credit assessment utilizing non-financial data; however, because such data originates from diverse sources and lacks a consistent structure, separate preprocessing and semantic extraction processes may be required. Furthermore, characteristics such as relationships, psychological signals, behavioral patterns, and lifestyle-based risk factors are difficult to capture with existing structured financial data; therefore, new data structures and modeling methods may be necessary to effectively apply these factors to credit assessment systems. The problem to be solved

[0005] The purpose of this specification to solve the aforementioned problems is to provide a technology that can more realistically evaluate an individual's creditworthiness by utilizing non-financial sector-based debt and credit information, lifestyle-based information, and relational information, which are not sufficiently reflected in the existing credit rating system calculated based on the transaction history of financial institutions.

[0006] Another objective of this specification to address the aforementioned problems is to improve the accuracy and explanatory power of credit assessment by structuring various non-financial data existing in different sources and formats, and analyzing credit-related behavioral signals, risk factors, and repayment propensity through an artificial intelligence-based model.

[0007] Another objective of this specification to solve the aforementioned problems is to provide technical means capable of calculating an individual's creditworthiness by integrally analyzing unstructured text, relationship-based information, behavior-based signals, and lifestyle-based attributes, thereby enabling a more balanced and explanatory credit assessment even for individuals with insufficient financial transaction history. means of solving the problem

[0008] A method for providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification for achieving the above objective may include: a step of acquiring source data for credit rating including information on the relationship between a creditor and a debtor, the purpose of the debt, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, nationality, whether the individual is staying overseas, the number of overdue payments, and information on occupation and age; a step of performing normalization and structuring processing on the source data, including text normalization based on natural language processing, rule-based parsing, outlier removal, and categorical encoding; a step of generating one or more derived variables representing repayment propensity, risk factors, lifestyle-based signals, and behavior-based signals from the normalized and structured data; a step of evaluating an individual's credit risk through an artificial intelligence agent that takes the derived variables and the normalized and structured data as input; and a step of calculating an individual's credit score or credit rating based on the evaluated credit risk.

[0009] Here, the server may include the step of generating a digital credit object that structures an individual's credit-related attributes, including at least some of the credit risk level, the derived variable, and the normalized and structured data.

[0010] Here, the server may include the steps of identifying question-and-answer-based interview content, consultation records, on-site verification results, and online text information included in source data for credit evaluation as unstructured text data; performing natural language processing-based text normalization, sentence separation, and part-of-speech analysis on the unstructured text data; and performing sentiment analysis, intent analysis, risk keyword detection, and contextual pattern analysis on the normalized unstructured text data to generate unstructured-based derived variables indicating an individual's repayment propensity, signals of responsibility, or potential for avoidance.

[0011] Here, the server may include the steps of: identifying the type of relationship, duration of the relationship, and level of mutual awareness between the creditor and the debtor included in the source data for credit evaluation as relationship-based information; calculating relationship characteristic values ​​indicating the intimacy of the relationship, the degree of information asymmetry, and the stability of the relationship based on the relationship-based information; and generating relationship-based derived variables indicating the debtor's willingness to repay, level of responsibility, or possibility of relationship severance based on the relationship characteristic values.

[0012] Here, the server may include the steps of: identifying the type of relationship between a creditor and a debtor included in source data for credit evaluation, the background of the formation of the relationship, and the duration of the relationship as relationship-based information; calculating an information accessibility index based on the relationship-based information regarding the possibility of a creditor accessing the debtor's occupation, income, residence, lifestyle patterns, or personal information, and evaluating the reliability of the information; and combining the relationship-based information and the reliability of the information to generate a relationship-information interaction-based derivative variable indicating information asymmetry, risk of relationship rupture, or possibility of misrecognition by relationship type.

[0013] Here, the server may include the steps of identifying the debtor's nationality, period of stay in Korea, and status of stay information included in source data for credit evaluation as nationality-based information; calculating a stability of stay index representing the possibility of long-term residence, possibility of movement abroad, and stability of stay in Korea based on the nationality-based information; and generating a nationality-based derivative variable representing the risk of long-term absence, possibility of contact, or change in credit risk due to departure based on the stability of stay index.

[0014] Here, the server may include the steps of: identifying the debtor's period of stay abroad, frequency of stay abroad, and recent departure status included in source data for credit evaluation as overseas stay-based information; calculating an absence risk index indicating the possibility of domestic absence, contactability, and potential transition to long-term overseas stay based on the overseas stay-based information; and generating an overseas stay-based derivative variable indicating the risk of loss of contact, the possibility of compliance with the repayment schedule, or changes in credit risk due to long-term absence based on the absence risk index.

[0015] Here, the server may include the steps of: identifying the debtor’s response rate relative to contact attempts, response delay time, and contact failure history included in the source data for credit evaluation as contact pattern-based information; calculating a contact reliability index that quantifies contact avoidance behavior, response consistency, and the degree of response bias by channel based on the contact pattern-based information; and generating behavior-based derived variables indicating the risk of contact failure, the possibility of compliance with the repayment schedule, or the possibility of responding to collection based on the contact reliability index.

[0016] Here, the server may include the steps of: collecting time series values ​​of risk factors derived from the above-mentioned derived variable or source data for credit evaluation to calculate the temporal change amount of the risk factor; analyzing the short-term volatility, medium-term trend, and abnormal fluctuation interval of the risk factor based on the temporal change amount of the risk factor; and generating a risk volatility-based derived variable indicating the possibility that credit risk may increase or decrease based on the analysis results.

[0017] An apparatus for providing a personal credit rating service based on non-financial bond data according to another embodiment of the present specification for achieving the above objective comprises a processor and a memory that stores at least one instruction executed by the processor. The at least one instruction acquires source data for credit rating including information on the relationship between a creditor and a debtor, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, nationality, whether the debtor resides overseas, the number of overdue payments, occupation, and age information. The instruction performs normalization and structuring processing on the source data for credit rating, including text normalization based on natural language processing, rule-based parsing, outlier removal, and categorical encoding. It generates one or more derived variables representing repayment propensity, risk factors, lifestyle-based signals, and behavior-based signals from the normalized and structured data. It evaluates an individual's credit risk through an artificial intelligence agent that takes the derived variables and the normalized and structured data as input, and may be configured to calculate an individual's credit score or credit rating based on the evaluated credit risk.

[0018] Here, the at least one command may be configured to generate a digital credit object that structures an individual's credit-related attributes, including at least some of the credit risk, the derived variable, and the normalized and structured data.

[0019] Here, the above at least one command may be configured to identify question-and-answer-based interview content, counseling records, on-site verification results, and online text information as unstructured text data, perform text normalization, sentence separation, and part-of-speech analysis on the unstructured text data, and perform sentiment analysis, intent analysis, risk keyword detection, and context pattern analysis on the normalized unstructured text data to generate unstructured-based derived variables.

[0020] Here, the at least one command may be configured to identify the type of relationship between the creditor and the debtor, the background of the formation of the relationship, and the duration of the relationship as relationship-based information, calculate the possibility of accessing debtor information based on the relationship-based information as an information accessibility index, evaluate information reliability according to the information accessibility index, and generate a relationship-information interaction-based derived variable by combining the relationship-based information and the information reliability.

[0021] Here, the at least one command may be configured to identify the debtor's nationality, period of stay in the country, and status of stay information as nationality-based information, calculate a stability of stay index indicating the possibility of long-term residence, possibility of movement abroad, and stability of stay in the country based on the nationality-based information, and generate nationality-based derived variables based on the stability of stay index.

[0022] Here, the at least one command may be configured to identify the debtor's period of stay abroad, frequency of stay abroad, and recent departure as overseas stay-based information, calculate an absence risk index indicating the possibility of domestic absence, possibility of contact, and possibility of transitioning to long-term overseas stay based on the overseas stay-based information, and generate an overseas stay-based derivative variable based on the absence risk index.

[0023] Here, the at least one command may be configured to identify the debtor's response rate relative to contact attempts, response delay time, and contact failure history as contact pattern-based information, calculate a contact reliability index including response consistency and channel response bias based on the contact pattern-based information, and generate behavior-based derived variables based on the contact reliability index.

[0024] Here, the at least one command may be configured to collect time series values ​​of risk factors derived from the derived variable or source data for credit evaluation to calculate a temporal change amount, analyze the short-term volatility, medium-term trend, and abnormal change interval of the risk factor based on the temporal change amount, and generate a risk volatility-based derived variable based on the analysis results. Effects of the invention

[0025] According to one embodiment of the present specification, various non-financial information can be reflected in the credit evaluation process even for individuals with insufficient financial transaction history, thereby mitigating the problem of information disparity that occurred in existing financial transaction-centered evaluation methods. In particular, although lifestyle-based information such as the relationship between the creditor and the debtor, the purpose of the debt, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, whether the individual is residing overseas, the number of overdue payments, occupation, and age is data that is difficult for financial institutions to obtain, it is likely to be closely related to the assessment of credit risk; therefore, by normalizing and structuring this information and reflecting it in the analysis, the explanatory power of the individual creditworthiness evaluation can be improved.

[0026] In addition, unstructured data obtained from interviews, Q&A-based surveys, on-site verification, etc., may include elements that are difficult to identify from structured data, such as the debtor's attitude, communication style, sense of responsibility, and behavioral signals related to repayment. According to the embodiments of this specification, since such unstructured information can be normalized based on natural language processing and converted into derived variables to be used as input for an artificial intelligence model, hidden risk factors or positive signals that were not reflected in existing credit rating systems can be included in the evaluation.

[0027] In addition, by comprehensively analyzing diverse non-financial data from different sources through AI-based models, multifaceted credit risk analysis becomes possible rather than a single-factor-centered evaluation. This can contribute to improving evaluation accuracy and prediction reliability by enabling creditworthiness predictions that reflect finer differences even for individuals with the same risk level. Brief explanation of the drawing

[0028] FIG. 1 is a system diagram including a server providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification. FIG. 2 is a block diagram showing the configuration of a server providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification. FIG. 3 is a flowchart illustrating a method for providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification. Specific details for implementing the invention

[0029] As the present specification is susceptible to various modifications and may have various embodiments, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present specification to specific embodiments, and it should be understood that it includes all modifications, equivalents, and substitutions that fall within the spirit and scope of the present specification. Similar reference numerals have been used for similar components in the description of each drawing.

[0030] Terms such as first, second, A, B, etc., may be used to describe various components, but said components should not be limited by said terms. These terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the rights of this specification, the first component may be named the second component, and similarly, the second component may be named the first component. The term "and / or" includes a combination of a plurality of related described items or any of a plurality of related described items.

[0031] When it is stated that one component is "connected" or "connected" to another component, it should be understood that while it may be directly connected or connected to that other component, there may also be other components in between. On the other hand, when it is stated that one component is "directly connected" or "directly connected" to another component, it should be understood that there are no other components in between.

[0032] The terms used in this application are used merely to describe specific embodiments and are not intended to limit this specification. The singular expression includes the plural expression unless the context clearly indicates otherwise. In this application, terms such as "comprising" or "having" are intended to indicate the presence of the features, numbers, steps, actions, components, parts, or combinations thereof described in the specification, and should be understood as not precluding the existence or addition of one or more other features, numbers, steps, actions, components, parts, or combinations thereof.

[0033] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as generally understood by those skilled in the art to which this specification pertains. Terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined in this application.

[0034] Hereinafter, preferred embodiments of the present specification will be described in more detail with reference to the attached drawings. To facilitate overall understanding in describing the present specification, the same reference numerals are used for identical components in the drawings, and redundant descriptions of identical components are omitted.

[0035] FIG. 1 is a system diagram including a server providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification.

[0036] Referring to FIG. 1, the method for providing a personal credit rating service based on non-financial bond data according to an embodiment of the present specification may be performed on a computing device that is equipped with storage space and connected to the Internet, such as a PC (Personal Computer), and is not easily portable, or on a portable terminal such as a smartphone. In this case, the method for providing a personal credit rating service based on non-financial bond data may be executed after an application implementing the method is downloaded from an App Store, etc., and installed on the portable terminal.

[0037] In addition, the method for providing a personal credit rating service based on non-financial bond data may be performed by inserting it into a computing device such as a PC while it is recorded on a recording medium such as a CD (Compact Disc) or USB (Universal Serial Bus) memory and executing it through an access operation of said computing device, or it may be performed by storing it from said recording medium into the storage space of said computing device and then executing it through an access operation of said computing device.

[0038] Meanwhile, if the above computing device or portable terminal can access a server connected to the Internet, the method for providing a personal credit rating service based on non-financial bond data can also be executed on the server upon a request from the computing device or portable terminal.

[0039] In the following, a computing device, portable terminal, or server, etc., on which the above method for providing a personal credit rating service based on non-financial bond data is executed may be collectively referred to as a device for providing a personal credit rating service based on non-financial bond data.

[0040] The above-mentioned device for providing personal credit rating services based on non-financial bond data may have the same configuration as the device for providing personal credit rating services based on non-financial bond data exemplified in FIG. 2, and the device for providing personal credit rating services based on non-financial bond data may not be limited to the device for providing personal credit rating services based on non-financial bond data illustrated in FIG. 1.

[0041] A system according to one embodiment may include a user terminal (110), a user terminal (120), a user terminal (130), and a server (140) providing a personal credit rating service based on non-financial bond data (hereinafter, server (140)).

[0042] The user terminal (110), user terminal (120), and user terminal (130) may be, but are not limited to, a vehicle, a mobility device, a smartphone, a tablet PC, a PC, a mobile phone, a PDA (personal digital assistant), a laptop, a media player, a micro server, a GPS (global positioning system) device, and other mobile or non-mobile computing devices. Additionally, the user terminal (110), user terminal (120), and user terminal (130) may be wearable devices equipped with communication functions and data processing functions. However, they are not limited to.

[0043] The server (140) may be implemented as a computer device or a plurality of computer devices that communicate with the user terminal (110), user terminal (120), and user terminal (130) through a network to provide commands, code, files, content, services, etc.

[0044] For example, the server (140) may provide a file for installing an application to a user terminal (110), a user terminal (120), and a user terminal (130) connected via a network. In this case, the user terminal (110), the user terminal (120), and the user terminal (130) may install the application using the file provided by the server (140).

[0045] Additionally, user terminals (110), user terminals (120), and user terminals (130) can connect to a server (140) under the control of an operating system (OS) and at least one program (e.g., a browser or an installed application) to receive services or content provided by the server (140).

[0046] As another example, the server (140) may establish a communication session for data transmission and reception and route data transmission and reception between user terminal (110), user terminal (120), and user terminal (130) through the established communication session.

[0047] User terminals (110), user terminals (120), user terminals (130), and a server (140) providing personal credit rating services based on non-financial bond data can communicate using a network. For example, the network includes a Local Area Network (LAN), a Wide Area Network (WAN), a Value Added Network (VAN), a mobile radio communication network, a satellite communication network, and combinations thereof, and is a data communication network in a comprehensive sense that enables each network constituent entity shown in FIG. 1 to communicate smoothly with one another, and may include wired internet, wireless internet, and mobile wireless communication networks. In addition, wireless communication may include, for example, wireless LAN (Wi-Fi), Bluetooth, Bluetooth Low Energy, LoRaWAN, Zigbee, WFD (Wi-Fi Direct), UWB (ultra wideband), infrared communication (IrDA, infrared Data Association), NFC (Near Field Communication), etc., but is not limited thereto.

[0048] FIG. 2 is a block diagram showing the configuration of a server providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification.

[0049] Referring to FIG. 2, a server (200) (hereinafter, server (200)) providing a personal credit rating service based on non-financial bond data may be configured to include a hardware processor (220), a communication unit (210), and a storage unit (230). Only components related to the embodiment are shown in the server (200) of FIG. 2. Therefore, a person skilled in the art will understand that other general-purpose components may be included in addition to the components shown in FIG. 2.

[0050] The communication unit (210) can operate to receive source data for credit evaluation by linking with an external terminal, a consultation system, a data input device, or an affiliated service platform. The communication unit (210) can be configured to support various protocols, such as HTTP-based API communication, message queue-based asynchronous data reception, and WebSocket-based real-time data reception, and can be designed to simultaneously process different data forms, such as debt relationship information, contact history, entry / exit status, period of stay, and Q&A records. Through this, the server (200) can receive multiple data streams containing a mixture of unstructured text data and structured data without difficulty.

[0051] The processor (220) can perform core operations for processing and analyzing received source data for credit evaluation by executing at least one command executed on the server (200). The processor (220) may be configured to include a multi-threaded computation engine, a text analysis module, a time series analysis module, a feature extraction unit, and an artificial intelligence agent execution environment. The processor (220) can perform normalization, tokenization, and intent analysis processes for text data, and can perform operations to produce various derived variables defined in the invention, such as relationship-based information, nationality-based information, information based on overseas stay, and contact-based behavioral signals. Additionally, the processor (220) can perform risk assessment by loading model parameters learned by the artificial intelligence agent, and produce credit prediction results through advanced computational operations such as risk volatility analysis.

[0052] The storage unit (230) may be implemented as a physical storage device for reliably storing data processed by the server (200). The storage unit (230) may be implemented using various storage technologies such as a relational database, a NoSQL database, a distributed file system, or object storage, and may store source data for credit evaluation, normalized data, derived variables, digital credit objects, and learning parameters of an artificial intelligence agent. The storage unit (230) may be configured to include a management module capable of managing data validity verification, error detection, version control, and model update history.

[0053] According to one embodiment of the present specification, the server (200) receives source data input through the communication unit (210), generates derivative variables for credit evaluation through the processor (220), calculates credit risk based on an artificial intelligence agent, and can store the result value in the storage unit (230) or link with other service modules. Accordingly, the server (200) is equipped with not only software analysis functions but also a physical computing environment and a storage environment to perform them, thereby enabling the entire process of the invention to be realistically implemented. In addition, the server (200) can be implemented in a cloud environment, an on-premises environment, or a hybrid environment, so it can respond to services of various scales.

[0054] The processor (220) may be implemented using at least one of ASICs (application specific integrated circuits), DSPs (digital signal processors), DSPDs (digital signal processing devices), PLDs (programmable logic devices), FPGAs (field programmable gate arrays), controllers, microcontrollers, microprocessors, and other electrical units for performing functions.

[0055] The storage unit (230) may be implemented as a non-transient computer-readable medium and may include both volatile memory (RAM) and non-volatile memory (ROM, flash memory, hard disk, etc.). The storage unit (230) may store the operating system, application programs, communication protocol stack, database, and learning model parameters of the artificial intelligence module of the server device. Additionally, the storage unit (230) may store log data, communication history, information received from an external terminal, etc., which can be utilized for subsequent operations.

[0056] The storage unit (230) may include RAM (random access memory), such as DRAM (dynamic random access memory) or SRAM (static random access memory), ROM (read-only memory), EEPROM (electrically erasable programmable read-only memory), CD-ROM, Blu-ray or other optical disc storage, HDD (hard disk drive), SSD (solid state drive), or flash memory. The storage unit (230) may also be referred to as memory.

[0057] The storage unit (230) may store at least one command executed through the processor (220). The at least one command may be configured to acquire source data for credit evaluation including information on the relationship between the creditor and the debtor, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, nationality, whether the individual is staying overseas, the number of overdue payments, occupation, and age information, perform normalization and structuring processing on the source data for credit evaluation including natural language processing-based text normalization, rule-based parsing, outlier removal, and categorical encoding, generate one or more derived variables representing repayment propensity, risk factors, lifestyle-based signals, and behavior-based signals from the normalized and structured data, evaluate an individual's credit risk through an artificial intelligence agent that takes the derived variables and the normalized and structured data as input, and calculate an individual's credit score or credit rating based on the evaluated credit risk.

[0058] Here, the at least one command may be configured to generate a digital credit object that structures an individual's credit-related attributes, including at least some of the credit risk, the derived variable, and the normalized and structured data.

[0059] Here, the above at least one command may be configured to identify question-and-answer-based interview content, counseling records, on-site verification results, and online text information as unstructured text data, perform text normalization, sentence separation, and part-of-speech analysis on the unstructured text data, and perform sentiment analysis, intent analysis, risk keyword detection, and context pattern analysis on the normalized unstructured text data to generate unstructured-based derived variables.

[0060] Here, the at least one command may be configured to identify the type of relationship between the creditor and the debtor, the background of the formation of the relationship, and the duration of the relationship as relationship-based information, calculate the possibility of accessing debtor information based on the relationship-based information as an information accessibility index, evaluate information reliability according to the information accessibility index, and generate a relationship-information interaction-based derived variable by combining the relationship-based information and the information reliability.

[0061] Here, the at least one command may be configured to identify the debtor's nationality, period of stay in the country, and status of stay information as nationality-based information, calculate a stability of stay index indicating the possibility of long-term residence, possibility of movement abroad, and stability of stay in the country based on the nationality-based information, and generate nationality-based derived variables based on the stability of stay index.

[0062] Here, the at least one command may be configured to identify the debtor's period of stay abroad, frequency of stay abroad, and recent departure as overseas stay-based information, calculate an absence risk index indicating the possibility of domestic absence, possibility of contact, and possibility of transitioning to long-term overseas stay based on the overseas stay-based information, and generate an overseas stay-based derivative variable based on the absence risk index.

[0063] Here, the at least one command may be configured to identify the debtor's response rate relative to contact attempts, response delay time, and contact failure history as contact pattern-based information, calculate a contact reliability index including response consistency and channel response bias based on the contact pattern-based information, and generate behavior-based derived variables based on the contact reliability index.

[0064] Here, the at least one command may be configured to collect time series values ​​of risk factors derived from the derived variable or source data for credit evaluation to calculate a temporal change amount, analyze the short-term volatility, medium-term trend, and abnormal change interval of the risk factor based on the temporal change amount, and generate a risk volatility-based derived variable based on the analysis results.

[0065] FIG. 3 is a flowchart illustrating a method for providing a personal credit rating service based on non-financial bond data according to one embodiment of the present specification.

[0066] Referring to FIG. 3, the server can obtain source data for credit evaluation (S300). For example, the server can obtain source data for credit evaluation including information on the relationship between the creditor and the debtor, the purpose of the debt, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, nationality, whether the debtor is staying overseas, the number of overdue payments, occupation, and age information.

[0067] According to one embodiment of the present specification, a server can perform a function for evaluating an individual's credit risk based on various forms of creditor and debtor-related data obtained in a non-financial environment. The server can receive source data for credit evaluation, including relationship information, debt-related Q&A, contact history, residence information, occupation information, and age information, from a user terminal, a consultation record input device, or an external service module through a communication unit. Since this source data is in a mixed form of structured and unstructured data, the server can perform preprocessing to convert it into a machine-processable structure before analyzing the data.

[0068] The server can perform normalization and structuring (S310). For example, the server can perform normalization and structuring processing on the source data, including natural language processing-based text normalization, rule-based parsing, outlier removal, and categorical encoding.

[0069] The server can perform normalization and structuring operations on received source data for credit evaluation, including text normalization, sentence splitting, part-of-speech analysis, rule-based parsing, outlier removal, and encoding of categorical data. For example, consultation records or online texts can be broken down by context to detect risk-related expressions, while structured data such as relationship information or occupational information can be converted into an analyzable form through unit-matched transformation. This preprocessing process can contribute to improving the accuracy of subsequent AI-based analysis by ensuring the formal consistency of the data.

[0070] The server can generate derived variables (S320). For example, the server can generate one or more derived variables representing repayment propensity, risk factors, life-based signals, and behavior-based signals from the normalized and structured data.

[0071] The server can generate derived variables based on normalized and structured data that reflect the characteristics of relationship-based information, residence-based information, behavioral contact signals, and debt-related questionnaires. These generated derived variables may include various risk factors not reflected in existing financial credit rating models, such as the debtor's repayment propensity, level of responsibility, risk of absence, stability of nationality and residence, potential for avoidance, and behavioral consistency. Such derived variables can be utilized as complex signals to enhance the accuracy of credit assessments.

[0072] The server can evaluate credit risk (S330). For example, the server can evaluate an individual's credit risk through an artificial intelligence agent that takes the above-derived variable and the above-described normalized and structured data as input.

[0073] The server can evaluate an individual's credit risk by executing an artificial intelligence agent that takes the aforementioned normalized data and derived variables as input. The AI ​​agent can analyze the probability of repayment failure, the intensity of risk signals, and gradual risk changes by utilizing learned model parameters, thereby calculating an overall credit risk level. The AI ​​agent can be designed to respond to imbalances or noise in various source data and can perform consistent risk calculations even when unstructured and structured data are input in combination.

[0074] The server can calculate a credit score (S340). For example, the server can calculate an individual's credit score based on the credit risk evaluated above.

[0075] The server can generate a final credit score or credit rating based on the risk level calculated by the artificial intelligence agent. The credit score can be calculated by comprehensively considering the influence of individual derived variables, the intensity of risk signals, and the volatility of risk; furthermore, it can be adjusted to reflect signals if specific risk factors surge or specific patterns persist. The calculated credit score may be provided to user terminals or external services and can be applied as a standard for evaluating an individual's creditworthiness in non-financial environments.

[0076] According to one embodiment of the present specification, a server can generate a digital credit object for structuring an individual's credit-related attributes based on an individual's credit risk and various derived variables generated from normalized data. The server can manage attributes corresponding to a specific user by grouping them into a single object form by synthesizing the credit risk calculated during the credit evaluation process, relationship-based information, residence-based information, behavior-based signals, and unstructured text analysis results.

[0077] Digital credit objects generated by the server may include a structure for integrally representing multiple data elements utilized in an individual's credit evaluation process. For example, credit risk may be expressed in the form of a score, relationship-based information may include attributes such as relationship type, duration, and information accessibility index, and residence-based information may include nationality, duration of residence, and residence stability index. Behavior-based signals may include attributes such as history of out-of-contact, response latency, and response consistency, and results of unstructured data analysis may include sentiment analysis indicators, the presence of risk keywords, or context-based risk patterns.

[0078] By structuring these attributes into objects, the server can store and manage a specific individual's credit profile in a consistent manner. Digital credit objects can be utilized for subsequent credit reassessment processes, risk tracking, volatility analysis, or credit verification functions based on user requests. Furthermore, digital credit objects can be updated when new data is collected or risk factors change, and the server can manage them by reflecting the latest state of the objects in the storage.

[0079] According to one embodiment of the present specification, a server can identify unstructured text data in sentence units among source data for credit evaluation, such as question-and-answer-based interview content, consultation records, on-site verification records, or online posts. Since such unstructured text data may contain various linguistic clues indicating a debtor's willingness to repay, living environment, financial pressure situation, or possibility of avoidance, the server can normalize the text and convert it into an analyzable form.

[0080] The server can perform text normalization, sentence splitting, tokenization, part-of-speech analysis, or sentence structure analysis on text data as a preprocessing step. For example, the accuracy of text analysis can be improved by normalizing expressions with the same meaning or removing unnecessary symbols within sentences. Additionally, the text separated into sentences can be fed into subsequent stages of sentiment analysis, intent analysis, or risk keyword detection.

[0081] The server can detect emotional signals, such as negative emotions, anxiety, avoidance psychology, or overly optimistic statements, by applying sentiment analysis algorithms to normalized unstructured text data. Additionally, through context-based analysis, it can identify expressions intended to delay repayment, evade responsibility, or indicate financial hardship. This analysis process can be performed by going beyond simple keyword detection to consider contextual meaning and linguistic patterns.

[0082] Based on these text analysis results, the server can generate unstructured-based derived variables indicating signals such as a lack of willingness to repay, potential for evasion, expressions of instability, or intentional exaggeration. These generated derived variables can complement behavioral and psychological factors that are difficult to capture with structured data during the credit risk assessment process. For example, if a debtor repeatedly uses evasive language during counseling, the server can quantify this pattern as a derived variable and incorporate it into the credit risk analysis.

[0083] According to one embodiment of the present specification, a server can convert relationship information and basic attribute information of a debtor included in source data for credit evaluation into structured data and structure it into an analyzable form. For example, since items such as debt amount, interest rate, time of contract conclusion, elapsed period, occupation classification, age group, or availability of contact may be recorded in different formats in the source data, the server can normalize them and convert them into a single standard structure.

[0084] The server can convert monetary information, time information, and categorical attributes usable in the credit evaluation process into numerical or categorical data tailored to their respective characteristic types. Numerical-based data, such as amounts or elapsed periods, can be standardized into an analyzable numerical form, while categorical data, such as occupations or relationship types, can be converted into a format usable by models based on the encoding method. This conversion process can help resolve formal differences in information provided by different data sources and contribute to reducing data inconsistencies that may occur during the analysis process.

[0085] The server can generate additional feature extensions or derived features during the structured data transformation process. For example, it can calculate the remaining term by utilizing elapsed time and maturity date information of a debt contract, or calculate an employment stability index based on job type. Furthermore, data such as whether there are recurring transactions between creditors and debtors or past repayment history can be structured in the form of binary or frequency values. This structured data-based processing can assist AI agents in effectively analyzing various data attributes.

[0086] According to one embodiment of the present specification, a server can identify relationship-based characteristics by analyzing relationship information between a creditor and a debtor included in source data for credit evaluation. The server can extract detailed attributes such as the type of relationship (family, acquaintance, workplace relationship, one-time transaction, etc.), the background of the formation of the relationship, the degree of mutual awareness, and the duration of the relationship, and such information can be utilized as an important factor reflecting the intimacy of the relationship, mutual trust, or the accessibility of information about the other party.

[0087] The server can estimate the likelihood of a creditor accessing a debtor's actual living information, occupation, income level, residence, or financial status based on the type of relationship. For example, in the case of a family relationship, it is highly likely that the creditor is aware of the debtor's occupation, lifestyle habits, or living situation with relative accuracy; in the case of a simple acquaintance relationship, they are likely to know only partial information; and in a one-time transaction relationship, information accessibility may be significantly low. The server can calculate an information accessibility index based on these relationship characteristics.

[0088] The server can evaluate the accuracy, omissions, and reliability of debtor-related information actually held by the creditor based on the calculated information accessibility index. The server can quantify information reliability by analyzing the veracity of occupational information included in the source data, the accuracy of address information, consistency with income levels, or the degree of omission in lifestyle information. For example, if the relationship is close but the provided information is inaccurate or contains many omissions, the server can reflect the discrepancy between relationship-based reliability and information-based reliability as a risk factor.

[0089] The server can generate relationship-information interaction-based derived variables by combining this relationship-based information with information reliability data. By analyzing the combination of relationship intimacy and information accuracy, the correlation between relationship duration and information asymmetry, or the linkage between the background of relationship formation and patterns of information omission, the server can express new types of risk, such as the risk of relationship severance, the possibility of information misinterpretation, or the possibility of avoiding repayment responsibility. These relationship-information interaction-based derived variables represent an area not reflected in existing financial credit rating models and can contribute to a more precise assessment of an individual's credit risk.

[0090] According to one embodiment of the present specification, a server can identify nationality-based characteristics by analyzing the nationality information of a debtor included in source data for credit evaluation. The server can use additional information, such as the duration of stay in the country, status of stay, visa type, conditions of permission to stay, or permanent residency status, in addition to nationality, to determine the stability of the debtor's stay in the country. This information can be used as factors reflecting the likelihood of the debtor staying in the country for a long period, the likelihood of long-term absence after departure, or a lifestyle pattern involving frequent movement between countries.

[0091] The server can calculate a residence stability index based on nationality-based information. The residence stability index is a value that quantifies the likelihood of continued residence in the country; it can be set to have a higher value when the duration of residence is longer, the status of residence is more stable, and the status is one authorized for long-term residence. Conversely, the residence stability index may be lower in cases of short-term residence status, frequent applications for extensions of residence, unstable visas, or short durations of residence. The server can calculate the debtor's domestic residence stability by combining these factors.

[0092] The server can generate nationality-based derived variables based on the calculated residence stability index. For example, if the residence stability index is low, the server can highlight risk factors such as the possibility of long-term absence, a potential transition to overseas residence, or reduced domestic contactability, and reflect these in the derived variables. Conversely, if the residence stability index is high, the residence stability-based risk may be assessed as low. The server can improve the accuracy of credit risk assessment by inputting the nationality-based derived variables as feature values ​​used by the AI ​​agent.

[0093] According to one embodiment of the present specification, a server can identify characteristics based on overseas stay by analyzing the debtor's overseas stay information included in source data for credit evaluation. The overseas stay information may include attributes such as the duration of overseas stay, frequency of overseas stay, whether there is a long-term departure, entry and exit records, and the time of the most recent departure. This information can be used as a signal regarding whether the debtor has been absent from the country for a certain period, whether overseas stay patterns appear repeatedly, or whether there is a possibility of long-term overseas movement.

[0094] Based on information regarding overseas residence, the server can calculate an absence risk index that quantifies the likelihood of domestic absence, the possibility of contact, and the potential transition to long-term overseas stay. For example, if records of recent departures appear consecutively or a pattern of prolonged overseas stays is detected, the server may be configured to increase the absence risk index, determining that the likelihood of the debtor being difficult to contact domestically has increased. Conversely, if there are no records of long-term stays or if entry and exit patterns are regular and stable, the absence risk index may be calculated as low.

[0095] The server can generate derivative variables based on overseas residence using the calculated absence risk index. For example, if the absence risk index is high, the server can reflect risk factors such as the likelihood of the debtor becoming unreachable, the possibility of repayment schedule delays, or reduced ability to respond to collection efforts as derivative variables. Conversely, if the absence risk index is low, the debtor's financial stability and repayment potential may be assessed as high. By inputting these derivative variables into the analysis of the AI ​​agent, the server can more accurately assess an individual's credit risk.

[0096] According to one embodiment of the present specification, a server can identify contact pattern-based information, such as the response rate relative to contact attempts, response delay time, and history of out-of-contact occurrences, included in source data for credit evaluation. This information can be used to determine how consistently a debtor responds to contact requests, whether they respond only to specific time periods or channels, or whether out-of-contact conditions occur repeatedly. The server can analyze contact-related behavioral signals by arranging these contact patterns in chronological order.

[0097] The server can calculate a contact reliability index that quantifies the debtor's contact avoidance behavior, response consistency, and the degree of response bias across channels, based on information derived from contact patterns. For example, the server can be configured to decrease the contact reliability index if response delays become prolonged or a history of missed contacts accumulates. Conversely, a high contact reliability index may be calculated if responses are stable within a consistent timeframe or if contact is made evenly across multiple channels. To calculate this contact reliability index, the server may utilize time-based statistical analysis, response pattern tracking, and behavior-based classification algorithms.

[0098] The server can generate behavior-based derived variables based on the calculated contact reliability index. For example, if the contact reliability index is low, risk factors such as the risk of the debtor becoming unreachable, reduced likelihood of adhering to repayment schedules, or decreased ability to respond to collection efforts can be reflected in the derived variables. By utilizing these behavior-based derived variables as input for the AI ​​agent, the server can assess an individual's credit risk more realistically. Furthermore, the server can determine whether the debtor's behavioral patterns are stable or deteriorating by analyzing the time-series changes in the contact reliability index.

[0099] According to one embodiment of the present specification, a server may collect time-series values ​​of risk factors derived from various derived variables generated during the credit evaluation process or source data for credit evaluation and analyze temporal changes. These time-series values ​​may include elements such as changes in relationship-based risk over a certain period, fluctuations in contact reliability, changes in overseas stay patterns, changes in the stay stability index, or changes in unstructured text-based risk signals. The server may structure the information to determine how each risk factor increases or decreases over time, or whether sudden surges or drops occur.

[0100] The server can analyze the short-term volatility, medium-term trends, and abnormal fluctuation periods of risk factors based on time series values. Short-term volatility can be calculated as an indicator of how rapidly a specific risk factor changes within a short period, while medium-term trends can be utilized as an indicator to determine the overall direction of the risk factor. Furthermore, the server can identify periods of abnormally large fluctuations as abnormal fluctuation periods by comparing them to usual patterns, and can additionally determine that the risk level is increasing if such periods repeat. To perform this volatility analysis, the server can apply time series analysis algorithms, moving average analysis, variance-based measurements, or anomaly detection algorithms.

[0101] The server can generate risk volatility-based derived variables based on the results of volatility analysis. For example, if the contact reliability index repeatedly declines over a short period, the server can be configured to increase the contact-based risk volatility variable. Additionally, if indicators related to overseas stay increase irregularly, the long-term absence risk volatility variable can be strengthened, and if relationship-based information becomes suddenly unstable, a relationship breakdown risk volatility variable can be generated. By utilizing these volatility-based derived variables as input values ​​for an AI agent, the server can perform a more detailed and sensitive credit risk assessment than existing static analysis-based methods.

[0102] According to one embodiment of the present specification, the server can capture signals of temporal fluctuations and abrupt changes in risk that are not considered in existing financial credit rating models by reflecting time-series changes in risk factors. This volatility-based analysis can contribute to enhancing the accuracy and realism of credit risk assessment by finely reflecting atypical risk signals related to an individual's actual life patterns.

[0103] The operation according to the embodiments of this specification may be implemented as a computer-readable program or code on a computer-readable recording medium. A computer-readable recording medium includes all types of recording devices in which data that can be read by a computer system is stored. Additionally, a computer-readable recording medium may be distributed across networked computer systems, allowing computer-readable programs or code to be stored and executed in a distributed manner.

[0104] When the embodiment is implemented in software, the above-described technique may be implemented as a module (process, function, etc.) that performs the above-described function. The module may be stored in memory and executed by a processor. The memory may be located inside or outside the processor and may be connected to the processor by various well-known means.

[0105] In addition, computer-readable recording media may include hardware devices specifically configured to store and execute program instructions, such as ROM, RAM, and flash memory. Program instructions may include machine code, such as that generated by a compiler, as well as high-level language code that can be executed by a computer using an interpreter, etc.

[0106] Some aspects of this specification have been described in the context of a device, but may also be described according to a corresponding method, wherein a block or device corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method may also be described according to a corresponding block or item or a feature of a corresponding device. Some or all of the method steps may be performed by (or using) a hardware device, such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, one or more of the most important method steps may be performed by such a device.

[0107] In the embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the methods described herein. In the embodiments, the field-programmable gate array may operate with a microprocessor to perform one of the methods described herein. Generally, it is preferable that the methods be performed by some hardware device.

[0108] Although the foregoing has been described with reference to preferred embodiments of this specification, those skilled in the art will understand that various modifications and changes can be made to this specification without departing from the spirit and scope of the specification as set forth in the following claims.

Claims

Claim 1 A method for providing a personal credit rating service based on non-financial bond data performed by at least one server, comprising: a step of acquiring source data for credit rating including information on the relationship between a creditor and a debtor, the purpose of the debt, the level of debtor information held by the creditor, whether contact is lost, whether personal rehabilitation or bankruptcy is in progress, nationality, whether the individual is staying overseas, the number of overdue payments, occupation, and age information; a step of performing normalization and structuring processing on the source data including natural language processing-based text normalization, rule-based parsing, outlier removal, and categorical encoding; a step of generating one or more derived variables representing repayment propensity, risk factors, lifestyle-based signals, and behavior-based signals from the normalized and structured data; a step of evaluating an individual's credit risk through an artificial intelligence agent that takes the derived variables and the normalized and structured data as input; and a step of calculating an individual's credit score based on the evaluated credit risk. Claim 2 A method for providing a personal credit rating service based on non-financial bond data according to claim 1, further comprising, after the step of calculating the credit score, the step of creating a digital credit object that structures the personal credit-related attributes including at least some of the credit risk level, the derived variable, and the normalized and structured data. Claim 3 A method for providing a personal credit rating service based on non-financial bond data according to claim 1, wherein the step of generating the derived variable comprises: identifying question-and-answer-based interview content, consultation records, on-site verification results, and online text information included in the source data for credit rating as unstructured text data; performing natural language processing-based text normalization, sentence separation, and part-of-speech analysis on the unstructured text data; and performing sentiment analysis, intent analysis, risk keyword detection, and contextual pattern analysis on the normalized unstructured text data to generate an unstructured-based derived variable indicating an individual's repayment propensity, responsibility signals, or avoidance potential. Claim 4 A method for providing a personal credit rating service based on non-financial bond data according to claim 1, wherein the step of generating the derived variable comprises: identifying as relationship-based information the type of relationship between a creditor and a debtor, the duration of the relationship, and the level of mutual awareness included in the source data for credit rating; calculating a relationship characteristic value indicating the intimacy of the relationship, the degree of unilateral information asymmetry, and the stability of the relationship with respect to the relationship-based information; and generating a relationship-based derived variable indicating the debtor's willingness to repay, the level of responsibility, or the possibility of relationship severance based on the relationship characteristic value. Claim 5 A method for providing a personal credit rating service based on non-financial bond data according to claim 1, wherein the step of generating the derivative variable comprises: identifying the type of relationship between a creditor and a debtor included in the source data for credit rating, the background of the formation of the relationship, and the duration of the relationship as relationship-based information; calculating an information accessibility index that indicates the possibility for a creditor to access the debtor's occupation, income, residence, lifestyle pattern, or personal information based on the relationship-based information; evaluating the accuracy, degree of omission, and reliability of the debtor-related information held by the creditor according to the information accessibility index; and generating a relationship information interaction-based derivative variable that indicates different information asymmetry, risk of relationship rupture, or possibility of misrecognition by relationship type by combining the relationship-based information and the evaluated information reliability.

Images