system

Abstract

We provide the system. [Solution] A means of inputting talk papers for high-expert crew and low-crew, A means for saving and pre-processing the input talk paper, A means for extracting keywords and phrases from a pre-processed talk paper, A means of comparing the frequency and context of extracted keywords and phrases and quantifying the discrepancies, A means of visualizing quantified deviation data and displaying the results, A system that includes means for formulating and managing training plans based on the displayed results.

Description

Technical Field

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

Background Art

[0002] Patent Document 1 discloses a method for controlling a persona chatbot, which is performed by at least one processor, including steps of receiving a user utterance, adding the user utterance to a prompt including an instruction sentence related to an explanation of a chatbot character, encoding the prompt, and inputting the encoded prompt into a language model to generate a chatbot utterance as a response to the user utterance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] Conventionally, it has been difficult to evaluate the quality of crew talk papers and quantify the performance differences between high - expert crews and low - level crews. As a result, there has been a problem that the training process for low - level crews is subjective and it is difficult to clarify specific improvement points. Also, formulating and managing a training plan has taken a lot of labor and time, and efficient training has not been achieved.

Means for Solving the Problems

[0005] This invention provides a means for inputting, saving, and pre-processing talk papers from high-expert crews and low-crews. Furthermore, the system includes means for extracting keywords and phrases from the pre-processed talk papers, comparing their frequency of occurrence and context, and quantifying the discrepancies. By analyzing the degree of discrepancy using a natural language processing algorithm and applying TF-IDF and cosine similarity, it is possible to obtain highly accurate results. By visualizing the quantified results and displaying them to the user, specific areas for improvement can be clarified, supporting the planning and management of efficient training programs.

[0006] Understood. Below are the definitions of the important words.

[0007] A "high-expert crew" is a crew member who possesses a high level of expertise and skills in their work and has an outstanding track record.

[0008] A "low-level crew" is a crew member who is still relatively inexperienced in their work and lacks the specialized knowledge and skills of a high-expert crew member.

[0009] A "talk paper" is a document that records the content of conversations and explanations used by crew members during their work.

[0010] "Preprocessing" is the process of preparing raw input data by converting its format and removing unnecessary parts, making it easier to analyze.

[0011] "Keywords" are words or phrases that have important meaning within the talk paper.

[0012] A "phrase" is a group of words or phrases that make sense within a talk paper.

[0013] "Frequency of occurrence" refers to the number of times a particular keyword or phrase appears within the talk paper.

[0014] "Context" refers to the context and situation before and after keywords or phrases are used.

[0015] "Deviation" refers to the difference in content and quality between the talk papers of the high-expert group and the low group.

[0016] "Quantification" refers to expressing qualitative data and information as numerical values.

[0017] "Visualization" refers to displaying quantified data and information in a visual form such as graphs or charts.

[0018] "Natural language processing algorithm" is a computational method for analyzing text data to understand the content and extract keywords.

[0019] "TF-IDF" is an abbreviation for Term Frequency-Inverse Document Frequency and is an index for evaluating how important a specific word is within a document.

[0020] "Cosine similarity" is a method for calculating the similarity of directions in vector space between two documents.

[0021] "Training plan" is a plan that sets specific goals and actions to improve the skills and knowledge of the low group.

Brief Explanation of Drawings

[0022] [Figure 1] It is a conceptual diagram showing an example of the configuration of a data processing system according to the first embodiment. [Figure 2] It is a conceptual diagram showing an example of the main functions of a data processing device and a smart device according to the first embodiment. [Figure 3] It is a conceptual diagram showing an example of the configuration of a data processing system according to the second embodiment. [Figure 4]It is a conceptual diagram showing an example of the main functions of a data processing device and smart glasses according to the second embodiment. [Figure 5] It is a conceptual diagram showing an example of the configuration of a data processing system according to the third embodiment. [Figure 6] It is a conceptual diagram showing an example of the main functions of a data processing device and a headset-type terminal according to the third embodiment. [Figure 7] It is a conceptual diagram showing an example of the configuration of a data processing system according to the fourth embodiment. [Figure 8] It is a conceptual diagram showing an example of the main functions of a data processing device and a robot according to the fourth embodiment. [Figure 9] It shows an emotion map to which a plurality of emotions are mapped. [Figure 10] It shows an emotion map to which a plurality of emotions are mapped. [Figure 11] It is a sequence diagram showing the processing flow of the data processing system in Example 1. [Figure 12] It is a sequence diagram showing the processing flow of the data processing system in Application Example 1. [Figure 13] It is a sequence diagram showing the processing flow of the data processing system in Example 2 when an emotion engine is combined. [Figure 14] It is a sequence diagram showing the processing flow of the data processing system in Application Example 2 when an emotion engine is combined.

Embodiments for Carrying Out the Invention

[0023] Hereinafter, an example of an embodiment of a system according to the technology of the present disclosure will be described with reference to the accompanying drawings.

[0024] First, the language used in the following description will be explained.

[0025] In the following embodiments, the signed processor (hereinafter simply referred to as "processor") may be a single arithmetic unit or a combination of multiple arithmetic units. Furthermore, the processor may be a single type of arithmetic unit or a combination of multiple types of arithmetic units. Examples of arithmetic units include CPU (Central Processing Unit), GPU (Graphics Processing Unit), GPGPU (General-Purpose computing on Graphics Processing Units), and APU (Accelerated Processing Unit).

[0026] In the following embodiments, signed RAM (Random Access Memory) is a memory that temporarily stores information and is used as work memory by the processor.

[0027] In the following embodiments, the signed storage is one or more non-volatile storage devices that store various programs and various parameters. Examples of non-volatile storage devices include flash memory (SSD (Solid State Drive)), magnetic disks (e.g., hard disks), or magnetic tapes.

[0028] In the following embodiments, the signed communication interface (I / F) is an interface that includes a communication processor and an antenna, etc. The communication interface manages communication between multiple computers. Examples of communication standards applicable to the communication interface include wireless communication standards such as 5G (5th Generation Mobile Communication System), Wi-Fi (registered trademark), or Bluetooth (registered trademark).

[0029] In the following embodiments, "A and / or B" is synonymous with "at least one of A and B." That is, "A and / or B" means that it may be A alone, or B alone, or a combination of A and B. Furthermore, in this specification, the same concept as "A and / or B" applies when expressing three or more things linked by "and / or."

[0030] [First Embodiment]

[0031] Figure 1 shows an example of the configuration of the data processing system 10 according to the first embodiment.

[0032] As shown in Figure 1, the 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.

[0033] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0034] The smart device 14 comprises a computer 36, a reception device 38, an output device 40, a camera 42, and a communication interface 44. The computer 36 comprises a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The reception device 38, output device 40, and camera 42 are also connected to the bus 52.

[0035] The reception device 38 is equipped with a touch panel 38A and a microphone 38B, etc., and receives user input. The touch panel 38A receives user input by detecting contact with an object (e.g., a pen or finger). The microphone 38B receives user input by detecting the user's voice. The control unit 46A transmits data indicating the user input received by the touch panel 38A and 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.

[0036] 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 perceptible to the user 20 (e.g., audio and / or text). The display 40A displays visible information such as text and images according to instructions from the processor 46. The speaker 40B outputs audio according to instructions from the processor 46. The camera 42 is a small digital camera equipped with an optical system such as a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor.

[0037] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various types of information between processor 46 and processor 28 via network 54.

[0038] Figure 2 shows an example of the main functions of the data processing device 12 and the smart device 14.

[0039] As shown in Figure 2, in the data processing device 12, a specific processing 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" related to the technology of this 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 according to the specific processing program 56 executed on the RAM 30.

[0040] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0041] In the smart device 14, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. The reception output program 60 is used in conjunction with a 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0042] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0043] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[0044] Program Processing Overview

[0045] 1. Data entry

[0046] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[0047] The device retrieves the uploaded talk paper and sends it to the server.

[0048] 2. Data Storage

[0049] The server saves the received talk papers to the database.

[0050] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[0051] 3. Text preprocessing

[0052] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization and stop word removal.

[0053] This formats the talk paper in a way that makes it easier to analyze.

[0054] 4. Extraction of keywords and phrases

[0055] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from the pre-processed talk paper.

[0056] The frequency of keyword occurrences is calculated, and this data is stored in a database.

[0057] 5. Data Comparison and Analysis

[0058] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews.

[0059] As a concrete example, we use TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[0060] Based on these analysis results, the gap between high-expert crews and low-expert crews will be quantified and evaluated.

[0061] 6. Visualization of Results

[0062] The server visualizes the quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a format that is easy for users to understand.

[0063] This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[0064] 7. Planning and management of training programs

[0065] Based on the results displayed by the system, users develop training plans for low-level crew members.

[0066] The server stores this training plan in a database and provides reminder and notification functions for progress management.

[0067] This allows for continuous tracking of progress towards achieving the training plan and enables adjustments to the plan as needed.

[0068] Specific example

[0069] Example 1:

[0070] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[0071] The server saves the talk paper to the database and performs preprocessing.

[0072] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[0073] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[0074] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[0075] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[0076] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[0077] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[0078] The following describes the processing flow.

[0079] Step 1:

[0080] The user accesses the system, enters their account information, and logs in.

[0081] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[0082] Step 2:

[0083] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[0084] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0085] Step 3:

[0086] The server saves the received talk papers to the database.

[0087] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[0088] Step 4:

[0089] The server passes the saved talk paper to the text analysis program.

[0090] The text analysis program performs preprocessing such as tokenization, stop word removal, and special character deletion.

[0091] Step 5:

[0092] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[0093] The frequency of occurrence of extracted keywords and phrases is calculated and stored in a database.

[0094] Step 6:

[0095] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0096] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[0097] Step 7:

[0098] The server visualizes the data in visual formats such as bar graphs and radar charts, based on the quantified deviation data.

[0099] Send the visualized data to the device.

[0100] Step 8:

[0101] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0102] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[0103] Step 9:

[0104] The user enters the training plan they have created into the system and sends it to the server.

[0105] The server saves the training plan to a database and sets up reminder and notification functions for progress management.

[0106] Step 10:

[0107] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[0108] (Example 1)

[0109] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0110] Conventional talk paper analysis systems have made it difficult to accurately grasp the performance differences between high-expert and low-level crews and to develop concrete training plans. Furthermore, there was a lack of systems to provide analysis results in a visually easy-to-understand format and to manage the progress of training plans based on those results. Therefore, the inability to achieve efficient and effective crew training remains a challenge.

[0111] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0112] In this invention, the server includes means for a user to log in and input data on high-expert crew and low-crew; means for a terminal to acquire the input data and send it to the server; means for the server to store the received data in a database and record metadata; means for the server to perform preprocessing on the stored data, such as tokenization and stop word removal; means for the server to extract keywords and important phrases from the preprocessed data; means for the server to compare the frequency of occurrence and usage of the extracted keywords and phrases and quantify the discrepancies; means for the server to visualize the quantified discrepancy data, create graphs and charts and send them to the terminal; means for the user to formulate a training plan based on the visualized results and save it to the server; and means for the server to manage the progress of the training plan and provide reminder and notification functions. This makes it possible to compare high-expert crew and low-crew, and to formulate and manage specific training plans based on the results.

[0113] A "user" refers to a person who accesses the system, enters data, or checks the results.

[0114] "Terminal" refers to a device used by a user, such as a computer or smartphone.

[0115] A "server" refers to a central processing unit that stores, processes, and analyzes data.

[0116] A "database" refers to a system for systematically storing and managing data.

[0117] "Metadata" refers to additional information accompanying the main data, such as the data type and creation date and time.

[0118] "Preprocessing" refers to a series of preparatory tasks performed before data analysis, such as tokenization and stop word removal.

[0119] "Tokenization" refers to the process of dividing text into smaller units such as words and phrases.

[0120] A "stop word" refers to a word that is frequently used in text analysis but does not have any particular meaning.

[0121] A "keyword" refers to a word or short phrase that has significant meaning within a text.

[0122] A "phrase" is a part of language, referring to a combination of multiple words that have a specific meaning on their own.

[0123] "Frequency" refers to an indicator of how often a particular word or phrase is used within a text.

[0124] "Discrepancy" refers to the differences in performance and terminology used between high-expert crews and low-level crews.

[0125] "Quantification" refers to representing data or results using specific numerical values.

[0126] "Visualization" refers to the visual representation of data using graphs and charts.

[0127] A "development plan" refers to a plan designed to improve the skills and performance of low-level crew members.

[0128] A "reminder" refers to an alert that is sent at a specific date or time.

[0129] "Notification function" refers to a system function that transmits information to the user.

[0130] Modes for carrying out the invention

[0131] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[0132] System Configuration

[0133] This system mainly consists of the following elements:

[0134] 1. User terminal

[0135] This is a device used by users to log in and upload talk papers. This includes personal computers, tablets, smartphones, etc.

[0136] 2. Server

[0137] It is a central processing unit that handles data storage, preprocessing, analysis, visualization, and management of development plans.

[0138] 3. Database

[0139] This is a storage device for saving talk papers, analysis results, and training plans. MongoDB will be used here.

[0140] Software to use

[0141] The software configuration of this system is mainly as follows:

[0142] 1. Text Analysis Library

[0143] NLTK (Natural Language Toolkit): Used for tokenizing text and removing stop words.

[0144] 2. Natural Language Processing (NLP) Libraries

[0145] SpaCy: Used to extract keywords and phrases from talk papers.

[0146] 3. Data Visualization Library

[0147] Matplotlib and Plotly are used to visualize quantified deviation data.

[0148] System operation

[0149] 1. Data entry

[0150] The user logs in and uploads the talk papers for the high-expert crew and the low-crew.

[0151] The terminal retrieves the talk paper and sends it to the server.

[0152] 2. Data Storage

[0153] The server saves the received talk papers to a database and records metadata (such as crew type and date / time).

[0154] 3. Text preprocessing

[0155] The server tokenizes the talk paper using NLTK and removes stop words.

[0156] 4. Extraction of keywords and phrases

[0157] The server uses SpaCy to extract keywords and important phrases from the pre-processed talk paper. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[0158] 5. Data Comparison and Analysis

[0159] The server compares the frequency of keywords and phrases used by high-expert crews and low-expert crews.

[0160] The server uses TF-IDF to calculate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[0161] 6. Visualization of Results

[0162] The server visualizes the quantified deviation data using Matplotlib or Plotly to create bar graphs and radar charts.

[0163] The device displays the visualized results to the user.

[0164] 7. Planning and management of training programs

[0165] Based on the results visualized by the user, a training plan for the low-level crew is developed and saved on the server.

[0166] The server manages the progress of the training plan and provides reminders and notifications.

[0167] Specific example

[0168] Example 1:

[0169] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[0170] The server saves the talk paper to the database and performs preprocessing.

[0171] The server uses NLP (Neuro-Linguistic Programming) techniques to extract keywords from the talk papers of Crew A and Crew B.

[0172] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[0173] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[0174] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[0175] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[0176] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[0177] Example of a prompt

[0178] "Analyze the following talk paper and quantify the performance gap between the high-expert crew and the low-expert crew."

[0179] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0180] Step 1:

[0181] The user logs into the system and uploads their talk papers for both the high-expert crew and the low-crew. The inputs here are the user's login information and the talk paper files. The terminal receives this information and sends it to the server. The login information is verified, and if successful, the talk paper upload screen appears. The user clicks the upload button, selects the file, and submits it.

[0182] Step 2:

[0183] The server saves the received talk paper to a database and records metadata (e.g., crew type, upload date and time). The input for this step is the talk paper file and metadata sent from the terminal. The server temporarily stores the file and converts it to a format suitable for storage in the database. It then saves it to the database along with the metadata. The output is the talk paper and metadata stored in the database.

[0184] Step 3:

[0185] The server preprocesses the stored talk papers using the NLTK library. The input for this step is the talk papers loaded from the database. Specifically, the server tokenizes the text (divides it into words) and removes stop words (e.g., "and," "the," etc.). This process formats the talk papers in a way that is suitable for analysis. The output is the preprocessed text data.

[0186] Step 4:

[0187] The server extracts keywords and important phrases from pre-processed text data using the SpaCy library. The input for this step is the pre-processed text data. The server identifies important phrases such as noun phrases and verb phrases and calculates their frequency. Specifically, the server calculates the frequency of occurrence of each keyword and stores it in the database. The output is data of the extracted keywords and their frequencies.

[0188] Step 5:

[0189] The server compares the frequency of keywords and phrases between high-expert and low-expert crews. The input for this step is keyword and phrase frequency data stored in a database. The server calculates TF-IDF to assess the importance of each keyword. It also uses cosine similarity to measure the similarity between talk papers. Specifically, the server quantifies the comparison results and stores them in a database. The output is quantified deviation data.

[0190] Step 6:

[0191] The server visualizes the quantified deviation data using Matplotlib or Plotly. The input for this step is the quantified deviation data. The server creates graphs in visually easy-to-understand formats such as bar graphs and radar charts. Specifically, the server generates graph data and sends it to the terminal. The output is the visualized graph data.

[0192] Step 7:

[0193] The terminal displays the visualized results to the user. The input for this step is the visualized graph data sent from the server. The terminal displays the graph on the screen so that the user can view it. Specifically, the graph is displayed in the terminal's web browser or similar application. The output is the graph displayed to the user.

[0194] Step 8:

[0195] The user develops a training plan for low-level crew members based on the visualized results. The input for this step is visualized graph data. The user fills in a specific training plan in an input form and sends it to the server. Specifically, the user decides on a training plan and enters it into the input form. The output is the training plan data sent to the server.

[0196] Step 9:

[0197] The server stores the training plan in a database and provides reminder and notification functions for progress management. The input for this step is the training plan data submitted by the user. The server stores the training plan in the database and sets reminders and notifications based on the schedule. Specifically, the server sets the notification schedule and sends reminders to the user. The output is the reminders and notifications sent to the user.

[0198] (Application Example 1)

[0199] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart device 14 will be referred to as the "terminal."

[0200] Traditionally, the efficiency improvements and training plans for robot operations within factories were primarily relied upon human experience and intuition. This resulted in insufficient transfer of knowledge and skills from experienced operators to new recruits, leading to decreased productivity and increased operational errors. Furthermore, the lack of concrete methods for incorporating the results of talk paper analysis into robot implementation made overall optimization difficult.

[0201] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0202] In this invention, the server includes means for inputting talk papers of high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to quantify the discrepancies between talk papers and efficiently plan and manage training programs. Furthermore, by linking it with a robot that optimizes operations, it is possible to immediately reflect the analysis results and improve the efficiency of the entire factory.

[0203] A "high-expert crew" is a group of operators who possess advanced expertise and skills in specific tasks or operations.

[0204] A "low crew" is a group of operators who are still in training or have low skill levels in a particular task or job.

[0205] A "talk paper" is a set of procedures, manuals, or customer service guidelines that operators refer to during their work.

[0206] "Input method" refers to the specific hardware or software interface used to input data into a system.

[0207] "Means of preservation" refer to storage devices and database systems for retaining data over the long term.

[0208] "Means of preprocessing" refers to the process of shaping and processing data to facilitate data analysis, and the program that executes it.

[0209] "Means for extracting keywords and phrases" refer to algorithms and techniques for identifying important words and phrases from text data.

[0210] "Methods for quantifying discrepancies" refer to methods and algorithms for measuring the differences between different datasets and expressing them numerically.

[0211] "Means of visualization and displaying results" refers to software or tools for displaying analyzed data in a visual form, such as graphs or charts.

[0212] "Means for formulating and managing training plans" refers to systems and methodologies for developing, tracking, and managing plans for improving operators' skills.

[0213] "Means for linking with a robot and displaying analysis results" refers to an interface and technology for integrating analyzed data into the control system of an industrial robot and displaying the results in real time.

[0214] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system is intended to optimize the operation of robots operating in a factory and is implemented in the following steps.

[0215] System Configuration

[0216] Data entry and saving:

[0217] The user first uploads the talk papers for the high-expert and low-expert crews to the system. The terminal retrieves these talk papers and sends them to the server. The server saves the received talk papers to its database. During this saving process, metadata such as the crew type and date and time are also recorded.

[0218] Text preprocessing and parsing:

[0219] The server first performs preprocessing on the talk papers stored in the database before analyzing them. This preprocessing includes tokenization and stop word removal. Through this process, the talk papers are formatted to be easily analyzed. Furthermore, natural language processing (NLP) techniques are used to extract keywords and important phrases from the preprocessed talk papers. Specific examples include using TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate keyword importance and using cosine similarity to measure the similarity between talk papers.

[0220] Data comparison and analysis:

[0221] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews. Based on this analysis, the discrepancy between high-expert and low-expert crews is quantified and evaluated. The quantified data is stored in a database.

[0222] Visualization and display:

[0223] The server visualizes quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a user-friendly format. This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[0224] Development and management of training plans:

[0225] Based on the results displayed by the system, users create training plans for their low-level crew members. The server stores these training plans in a database and provides reminders and notifications for progress management. This allows for continuous tracking of the training plan's progress and adjustments to the plan as needed.

[0226] Robot collaboration and optimization:

[0227] The server transmits the analysis results to the industrial robots. This allows for real-time display of the analysis results and optimization of operations. This integration supports the overall efficiency of the factory.

[0228] Specific example

[0229] For example, if a high-expert crew uses a talk paper that emphasizes "building trust for customer satisfaction and quick response," and a low-crew uses a talk paper that emphasizes "building trust with customers for quick response," the server compares these talk papers. First, it extracts keywords from each talk paper and calculates the TF-IDF value. Then, it calculates cosine similarity to quantify the divergence between the two talk papers. The quantified data is visualized in bar graphs or radar charts and presented to the user.

[0230] Examples of prompts for generative AI models

[0231] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[0232] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0233] Step 1:

[0234] The user logs into the system and uploads talk papers for both the high-expert crew and the low-crew. The terminal retrieves these talk papers and sends them to the server. The talk papers are expected to be in file formats such as text files or PDFs.

[0235] Input: Talk papers for the high-expert crew and the low-crew crew (text or PDF format).

[0236] Output: Talk paper sent to the server.

[0237] Step 2:

[0238] The server saves uploaded talk papers to the database. Along with the talk papers, metadata (such as crew type and upload date / time) is also saved. This metadata will be useful for future data analysis.

[0239] Input: Talk paper and metadata sent to the server.

[0240] Output: Talk papers and metadata stored in the database.

[0241] Step 3:

[0242] The server retrieves the talk papers stored in the database for preprocessing. Preprocessing includes tokenization, stop word removal, and normalization. These preprocessing steps prepare the text data for easier analysis.

[0243] Input: Talk paper retrieved from the database.

[0244] Output: Preprocessed text data.

[0245] Step 4:

[0246] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from pre-processed talk papers. Specifically, TF-IDF (Term Frequency-Inverse Document Frequency) is used.

[0247] Input: Preprocessed text data.

[0248] Output: Extracted keywords and phrases, and their frequency data.

[0249] Step 5:

[0250] The server uses TF-IDF to evaluate the importance of each keyword and then uses cosine similarity to measure the similarity between talk papers from high-expert and low-expert crews. This analysis quantifies the discrepancies between talk papers.

[0251] Input: Keywords and phrases, and their frequency data.

[0252] Output: Numerical deviation data.

[0253] Step 6:

[0254] The server visualizes quantified deviation data. It creates bar graphs and radar charts, displaying the results in a user-friendly format. This allows users to visually identify areas that require improvement.

[0255] Input: Numerical deviation data.

[0256] Output: Results visualized as bar graphs or radar charts.

[0257] Step 7:

[0258] The user develops a training plan for the low-level crew based on the results displayed within the system. The server stores this training plan in a database and provides reminders and notifications for progress management.

[0259] Input: Visualized result.

[0260] Output: Training plan stored in the database.

[0261] Step 8:

[0262] The server links the analysis results to the control system of the industrial robot. This linkage allows the analysis results to be reflected in the robot's operation in real time, optimizing the work process.

[0263] Input: Analysis results.

[0264] Output: Analysis results reflected in the robot.

[0265] Specific examples of operation

[0266] If a senior operator's talk paper states "We prioritize trust for customer satisfaction and prompt response," and a junior operator's talk paper states "We build trust with customers to respond quickly," the server analyzes these talk papers. It extracts keywords, compares their frequencies, and derives several characteristics. Then, it quantifies the discrepancies, visualizes them, and provides them to the user. Based on these results, the user creates a training plan, which the server saves to a database. Finally, the analysis results are linked to a robot to optimize operations.

[0267] Examples of prompts for generative AI models

[0268] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[0269] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0270] This invention combines a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies with an emotion engine. This system not only analyzes the talk papers entered by users and quantifies the discrepancies, but also recognizes and analyzes the user's emotions and reflects them in the training plan, thereby supporting more effective training.

[0271] Program Processing Overview

[0272] 1. Data Input

[0273] The user logs in to the system and uploads the talk papers of the high-expert group and the low group.

[0274] The terminal obtains the uploaded talk papers as text data and sends them to the server.

[0275] 2. Data Storage

[0276] The server saves the received talk papers in the database.

[0277] Metadata (such as the type of group, date and time, etc.) is also recorded when saving the talk papers.

[0278] 3. Text Pretreatment

[0279] The server passes the talk papers to a text analysis program for pretreatment such as tokenization, stop word removal, and special character deletion.

[0280] 4. Extraction of Keywords and Phrases

[0281] The server uses natural language processing (NLP) technology to extract important keywords and phrases from the pretreated talk papers.

[0282] Calculate the frequency of occurrence of the extracted keywords and phrases and save the data in the database.

[0283] 5. Data Comparison and Analysis

[0284] The server evaluates the importance of keywords for the talk papers of the high-expert group and the low group using the TF-IDF algorithm.

[0285] Calculate the similarity between the two talk papers using cosine similarity and quantify the divergence.

[0286] 6. Application of the Emotion Engine

[0287] The server uses the emotion engine to analyze the user's emotion from the keywords and phrases in the talk paper.

[0288] Classify the user's emotion into positive, negative, neutral, etc., and save the result in the database.

[0289] 7. Visualization of Results

[0290] The server visualizes the quantified divergence data and emotion data. For example, in addition to bar graphs and radar charts, display indicators related to emotion.

[0291] Send the visualized data to the terminal.

[0292] 8. Formulation and Management of Training Plans

[0293] Based on the visualized results, the user formulates a training plan for the low crew.

[0294] The server automatically adjusts the content of the training plan by referring to the emotion data. For example, if there is a lot of negative emotion, incorporate elements of mental support.

[0295] The server saves the training plan in the database and provides reminder and notification functions for progress management.

[0296] Specific Example

[0297] Example 1:

[0298] The user logs in to the system and uploads the talk papers of high expert crew A and low crew B.

[0299] The server saves the talk papers in the database and performs preprocessing.

[0300] The server uses NLP technology to extract respective keywords (such as "customer satisfaction", "prompt response", "trust relationship") from the talk papers of Group A and Group B.

[0301] The server uses TF-IDF and cosine similarity to quantify and evaluate the divergence between the talk papers of Group A and Group B.

[0302] The server uses an emotion engine to analyze the user's emotions from the keywords and phrases in the talk papers. For example, if negative keywords such as "difficulty" and "failure" frequently appear in the talk papers of Group B, the user's emotion is classified as "negative".

[0303] The server visualizes the quantified data and emotion data in a bar graph or radar chart, and also displays indicators related to emotions.

[0304] The terminal displays the visualized results to the user, and clarifies specific improvement points and areas that need cultivation.

[0305] Based on the results displayed to the user, the user formulates a cultivation plan for Low Group B and saves the plan in the server.

[0306] The server automatically adjusts the cultivation plan by referring to the emotion data, and adds elements such as mental training and counseling, for example.

[0307] The server supports the progress management of the cultivation plan and sends necessary reminders and notifications to the user.

[0308] As a result, it becomes possible to objectively evaluate the performance difference between the high-expert group and the low group, and to formulate a specific cultivation plan considering the user's emotions.

[0309] The following describes the processing flow.

[0310] Step 1:

[0311] The user accesses the system, enters their account information, and logs in.

[0312] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[0313] Step 2:

[0314] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[0315] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0316] Step 3:

[0317] The server saves the received talk papers to the database.

[0318] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[0319] Step 4:

[0320] The server passes the saved talk paper to the text analysis program.

[0321] The text analysis program performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[0322] Step 5:

[0323] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[0324] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[0325] Step 6:

[0326] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0327] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[0328] Step 7:

[0329] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[0330] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[0331] Step 8:

[0332] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[0333] Send the visualized data to the device.

[0334] Step 9:

[0335] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0336] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[0337] Step 10:

[0338] The user enters the training plan they have created into the system and sends it to the server.

[0339] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[0340] Step 11:

[0341] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[0342] Step 12:

[0343] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[0344] (Example 2)

[0345] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the smart device 14 as the "terminal".

[0346] Conventional training planning systems made it difficult to compare the performance of high-expert crew members with that of low-level crew members, and they were unable to create specific training plans that took into account user emotions. Furthermore, there was a lack of means to quantify discrepancies in text documents and analyze and visualize emotional data, which hindered effective training support.

[0347] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0348] In this invention, the server includes means for inputting text documents of high-expert crew and low-crew; means for saving and pre-processing the input text documents; means for extracting keywords and phrases from the pre-processed text documents; means for comparing the frequency and context of the extracted keywords and phrases and quantifying the discrepancies; means for analyzing sentiment from the keywords and phrases and saving the results; means for visualizing the quantified discrepancy data and sentiment data and displaying the results; and means for formulating and managing training plans based on the displayed results. This makes it possible to objectively evaluate the performance differences between high-expert crew and low-crew, formulate specific training plans that take into account user sentiment, and provide effective training support.

[0349] A "text document" refers to a document file saved in text format, including, for example, PDF, Word, and Excel files.

[0350] "Input method" refers to the interface that allows users to upload text documents to the system, and it uses specific software and hardware.

[0351] "Means of saving and pre-processing" refers to the process by which the server saves the received text document to a database and then performs text processing such as tokenization, stop word removal, and removal of special characters.

[0352] "Means for extracting keywords and phrases" refers to the process of identifying and extracting important words and phrases from a text document using natural language processing algorithms.

[0353] "Methods for comparing frequency of occurrence and context and quantifying the discrepancies" refers to the process of analyzing the frequency of occurrence and contextual information of extracted keywords and phrases, and expressing the differences between high-expert crews and low-level crews numerically.

[0354] "Methods for analyzing emotions and saving results" refers to the process of using an emotion engine to identify a user's emotions from keywords and phrases within a text document and recording the results in a database.

[0355] "Means of visualization and displaying results" refers to an interface that visually represents quantified deviation data and sentiment data in the form of graphs and charts, and displays them to the user.

[0356] "Means for formulating and managing training plans" refers to the function of formulating training policies for low-level employees based on visualized data, and tracking and managing their progress.

[0357] This invention combines a system that analyzes text documents (hereinafter referred to as "talk papers") of high-expert crews and low-level crews and quantifies the discrepancies between them with an emotion engine. This system aims to support more effective training by not only analyzing the talk papers entered by users and quantifying the discrepancies, but also recognizing and analyzing the user's emotions and reflecting them in the training plan.

[0358] Hardware and software to be used

[0359] The system consists of the following main hardware and software components.

[0360] Device: A device such as a PC, smartphone, or tablet used by the user to upload their talk paper. Web applications or mobile applications are used.

[0361] Server: A central processing unit for data storage, preprocessing, analysis, and visualization. Server software used includes Apache® and Nginx, among others.

[0362] Database: A relational database management system (RDBMS) used to store talk papers and analysis results. MySQL® and PostgreSQL are commonly used.

[0363] NLP libraries: Software libraries for natural language processing. These include NLTK, SpaCy, BERT, Word2Vec, and others.

[0364] Emotion engine: APIs or software used for sentiment analysis. Examples include Google Cloud Natural Language API and IBM Watson.

[0365] Graph libraries: Tools for visualizing data. D3.js and Chart.js are commonly used.

[0366] Specific processing flow

[0367] 1. Data entry

[0368] Users log into the system and upload the talk papers for their high-expert and low-expert crews. For example, a user selects the talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf) from their PC or smartphone and clicks the upload button.

[0369] 2. Data storage and preprocessing

[0370] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[0371] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[0372] 3. Keyword and phrase extraction

[0373] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from pre-processed talk papers. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[0374] 4. Data Comparison and Analysis

[0375] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[0376] The server uses the TF-IDF algorithm to evaluate the importance of keywords, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[0377] 5. Application of the Emotion Engine

[0378] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" appear frequently, the user's emotion will be classified as "negative."

[0379] 6. Visualization of Results

[0380] The server visualizes quantified deviation and sentiment data. In addition to bar graphs and radar charts, it displays sentiment-related indicators.

[0381] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0382] 7. Planning and management of training programs

[0383] Based on the results visualized by the user, a training plan for low-level crew members will be developed. For example, "mental training courses" and "communication skills improvement sessions" will be considered.

[0384] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[0385] A possible example of a specific prompt might be, "Extract keywords from the talk papers of the high-expert crew and low-expert crew, analyze their emotions, and incorporate them into the training plan."

[0386] This series of processes makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, develop specific training plans that take user emotions into consideration, and provide effective training support.

[0387] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0388] Step 1:

[0389] Data entry

[0390] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[0391] Input: Talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf)

[0392] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0393] Output: Text data sent to the server

[0394] Specific actions:

[0395] The user selects a talk paper from their PC or smartphone and clicks the upload button. The device reads the file and sends it to the server as text data.

[0396] Step 2:

[0397] Data storage and preprocessing

[0398] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[0399] Input: Text data sent to the server

[0400] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[0401] Output: Preprocessed text data

[0402] Specific actions:

[0403] The server saves the talk paper to the database, recording metadata such as the file name, upload date and time, and user ID. Next, the server uses a text analysis program to perform tokenization, stop word removal, and special character removal to generate pre-processed text data.

[0404] Step 3:

[0405] Keyword and phrase extraction

[0406] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from the pre-processed talk paper.

[0407] Input: Preprocessed text data

[0408] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[0409] Output: List of keywords and phrases and their frequencies

[0410] Specific actions:

[0411] The server applies natural language processing technology to extract important keywords (e.g., "customer satisfaction," "prompt response") and phrases from the talk paper. It calculates the frequency of occurrence of the extracted keywords and stores the results in a database.

[0412] Step 4:

[0413] Data comparison and analysis

[0414] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0415] Input: A list of keywords and phrases and their frequency of occurrence.

[0416] The server uses cosine similarity to calculate the similarity between the two talk papers and quantifies the discrepancy.

[0417] Output: Quantified deviation data

[0418] Specific actions:

[0419] The server applies the TF-IDF algorithm to calculate the importance of keywords within the talk paper. Next, it calculates cosine similarity to quantify the discrepancy between the talk papers of the high-expert crew and the low-crew crew.

[0420] Step 5:

[0421] Application of the emotion engine

[0422] The server uses an emotion engine (e.g., Google Cloud Natural Language API or IBM Watson) to analyze the user's emotions from keywords and phrases within the talk paper.

[0423] Input: Preprocessed text data

[0424] The server classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[0425] Output: Sentiment analysis result data

[0426] Specific actions:

[0427] The server uses an emotion engine to analyze keywords and phrases within the talk paper and determine an emotion score and emotion type. The results are then stored in a database.

[0428] Step 6:

[0429] Visualization of results

[0430] The server visualizes quantified deviation data and sentiment data.

[0431] Input: Quantified deviation data and sentiment analysis result data

[0432] The server uses a graphing library (e.g., D3.js or Chart.js) to visualize this data in bar graphs or radar charts.

[0433] Output: Visualized graphs and charts

[0434] Specific actions:

[0435] The server generates graphs based on deviation data and sentiment data, prepares data for display in a visually easy-to-understand format, and sends this data to the terminal.

[0436] Step 7:

[0437] Development and management of training plans

[0438] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[0439] Input: Visualized graphs and charts

[0440] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[0441] Output: Training plan and progress management data

[0442] Specific actions:

[0443] The user views the displayed graph data, creates a specific training plan for the low-level crew member, and saves it to the server. The server then takes sentiment data into consideration and automatically adjusts the training plan, setting up relevant reminders and notifications.

[0444] (Application Example 2)

[0445] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as a "server" and the smart device 14 as a "terminal".

[0446] Previous methods for evaluating staff performance and developing training plans relied heavily on subjective assessments, making effective training difficult. Furthermore, failing to consider staff emotional states during training planning could lead to decreased motivation and stress. This invention aims to provide more effective training plans by quantifying the discrepancies between high-expert and low-level staff communication reports and by considering staff emotional states.

[0447] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0448] In this invention, the server includes means for inputting talk papers from high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to visualize quantified discrepancy data and sentiment data, formulate and manage training plans based on the results, and periodically monitor and notify progress. Furthermore, by including means for analyzing the user's emotions from keywords and phrases in the talk papers using an emotion analysis engine, it becomes possible to formulate training plans that reflect the emotional state of the staff.

[0449] A "high-expert crew" is a staff member who possesses high levels of experience and skills and demonstrates outstanding performance in their work.

[0450] A "low-level crew" refers to staff members who have relatively low experience and skills, and who still have room for improvement in their work.

[0451] A "talk paper" refers to a document or script that staff members use when responding to customers or giving explanations.

[0452] "Means of input" refers to devices or interfaces used to import staff members' talk papers into the system.

[0453] "Means of storage" refers to storage or databases used to record and retain entered data.

[0454] "Methods for preprocessing" refer to processes such as text tokenization, stop word removal, and special character deletion performed to prepare the input data into a format that is easy to analyze.

[0455] "Means for extracting keywords and phrases" refers to methods for extracting important words and phrases using natural language processing algorithms.

[0456] "Methods for quantifying discrepancies" refer to algorithms and analytical methods for quantitatively evaluating and expressing the differences between talk papers from high-expert crews and low-expert crews as numerical values.

[0457] "Means of visualization" refer to tools and software used to display analysis results in graphical formats such as bar graphs and radar charts.

[0458] "Means for formulating and managing training plans" refers to systems and mechanisms for formulating staff training plans, monitoring their progress, and issuing necessary notifications and reminders.

[0459] An "emotion analysis engine" refers to algorithms and technologies that read emotions from text data and classify them as positive, negative, neutral, etc.

[0460] "Monitoring and notification mechanisms" refer to functions that allow the system to periodically check the progress of the training plan and generate reminders and alerts as needed.

[0461] This invention consists of a system for inputting talk papers from high-expert crew members and low-level crew members, quantifying the discrepancies between them, and formulating and managing staff training plans. In addition, by using an emotion analysis engine to analyze the user's emotions and reflecting them in the training plan, it is possible to support more effective training.

[0462] Hardware and software to be used

[0463] Hardware:

[0464] Server: Used for processing and storing data.

[0465] Device: An input device such as a smartphone or smart glasses.

[0466] software:

[0467] Speech recognition software (Google Speech Recognition): Converts spoken conversation into text data.

[0468] Natural language processing libraries (nltk, scikit-learn): Used for text tokenization, stop word removal, special character removal, and keyword extraction.

[0469] Database: Used to store talk papers and analysis results.

[0470] Emotion analysis engine: Recognizes and analyzes emotions such as positive, negative, and neutral.

[0471] Program Processing Overview

[0472] Data entry and saving

[0473] Users upload talk papers for high-expert and low-level crews to the system using their smartphones or smart glasses. The server receives this data and stores it in a database. At the same time, it also records metadata (crew type, date and time, etc.).

[0474] Text preprocessing

[0475] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and special character deletion. This prepares the text for easier analysis.

[0476] Keyword and phrase extraction and comparison

[0477] The server uses natural language processing techniques to extract important keywords and phrases from pre-processed talk papers. Then, it evaluates the importance of the keywords using the TF-IDF algorithm, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[0478] Emotion analysis

[0479] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases in the talk paper. It classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[0480] Visualization of results

[0481] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts. This visualized data is sent to the terminal, and the results are displayed to the user.

[0482] Development and management of training plans

[0483] Based on the user's visualized results, the server develops a training plan for the low-crew member. The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it incorporates elements of mental support. The server supports the progress management of the training plan and provides necessary reminders and notifications.

[0484] Specific examples of implementation

[0485] For example, imagine a system where smart glasses are used to record conversations between staff working in a physical store and customers, and the data is analyzed. The recorded conversations are sent to a server and compared to the talk papers of high-expert crew members. Based on the analysis results, a staff training plan is developed. The progress of the plan is then monitored regularly, and notifications are sent to the user as needed. If the emotional data is negative, mental training is suggested.

[0486] Example of a prompt

[0487] "Please analyze the conversation text, calculate its similarity to advanced customer service skills, and perform sentiment analysis. The following conversation text: {Conversation Text}"

[0488] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0489] Step 1:

[0490] Data entry

[0491] Users record the talk papers of high-expert and low-crew members using their smartphones or smart glasses and upload them to the system. The device converts the recorded audio data into text data and sends it to the server.

[0492] Input: Audio data

[0493] Output: Text data

[0494] Specific operation: Use speech recognition software (Google Speech Recognition) to convert speech data into text.

[0495] Step 2:

[0496] Data storage

[0497] The server stores the received text data in a database. At the same time, it also records metadata related to the talk paper (such as crew type and date / time).

[0498] Input: Text data, metadata

[0499] Output: Saved database entries

[0500] Specific operation: Connect to the database and save text data and metadata in a specified format.

[0501] Step 3:

[0502] Pre-treatment

[0503] The server retrieves talk papers from the database and performs preprocessing. Specifically, it performs tokenization, stop word removal, and removal of special characters.

[0504] Input: Text data

[0505] Output: Preprocessed text data

[0506] Specific operation: Uses the natural language processing library (nltk) to perform tokenization, stop word removal, and special character removal.

[0507] Step 4:

[0508] Keyword and phrase extraction

[0509] The server extracts important keywords and phrases from the pre-processed text data.

[0510] Input: Preprocessed text data

[0511] Output: List of keywords and phrases

[0512] Specific operation: Use TfidfVectorizer to evaluate and extract keywords and phrases from text based on their importance.

[0513] Step 5:

[0514] Quantifying the discrepancy

[0515] The server uses the TF-IDF algorithm and cosine similarity to calculate the similarity between the talk papers of the high-expert crew and the low-crew, and quantifies the discrepancy.

[0516] Input: List of keywords and phrases

[0517] Output: Quantified deviation data

[0518] Specific operation: Calculates similarity using TfidfVectorizer and cosine_similarity.

[0519] Step 6:

[0520] Emotion analysis

[0521] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases within the talk paper.

[0522] Input: List of keywords and phrases

[0523] Output: Sentiment analysis results (positive, negative, neutral, etc.)

[0524] Specific operation: Classifies the sentiment of keywords and phrases using a proprietary sentiment analysis algorithm.

[0525] Step 7:

[0526] Visualization of results

[0527] The server visualizes quantified deviation data and sentiment data and sends the results to the terminal.

[0528] Input: Quantified deviation data, sentiment data

[0529] Output: Visualized data (graphs, charts, etc.)

[0530] Specific operation: Use matplotlib to generate and visualize bar graphs and radar charts.

[0531] Step 8:

[0532] Development and management of training plans

[0533] Users create training plans for their low-level crew members based on visualized results. The server automatically adjusts the training plan based on sentiment data, monitors progress, and provides notifications.

[0534] Input: Visualized data, training plan

[0535] Output: Adjusted training plan, progress notifications

[0536] Specific operation: The server adjusts the care plan based on sentiment data stored in the database, monitors progress, and generates reminders and notifications as needed.

[0537] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating 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.

[0538] Data generation model 58 is a so-called generative AI (Artificial Intelligence). An example of data generation model 58 is ChatGPT (registered trademark) (Internet search).<URL: https: / / openai.com / blog / chatgpt> ), Gemini (registered trademark) (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[0539] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart device 14.

[0540] [Second Embodiment]

[0541] Figure 3 shows an example of the configuration of the data processing system 210 according to the second embodiment.

[0542] As shown in Figure 3, the data processing system 210 includes a data processing device 12 and smart glasses 214. An example of the data processing device 12 is a server.

[0543] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[0544] The smart glasses 214 include a computer 36, a microphone 238, a speaker 240, a camera 42, and a communication interface 44. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, and camera 42 are also connected to the bus 52.

[0545] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[0546] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[0547] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[0548] Figure 4 shows an example of the main functions of the data processing device 12 and the smart glasses 214. As shown in Figure 4, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[0549] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[0550] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[0551] In the smart glasses 214, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[0552] Next, the identification processing performed by the identification processing unit 290 of the data processing device 12 will be described. 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".

[0553] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[0554] Program Processing Overview

[0555] 1. Data entry

[0556] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[0557] The device retrieves the uploaded talk paper and sends it to the server.

[0558] 2. Data Storage

[0559] The server saves the received talk papers to the database.

[0560] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[0561] 3. Text preprocessing

[0562] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization and stop word removal.

[0563] This formats the talk paper in a way that makes it easier to analyze.

[0564] 4. Extraction of keywords and phrases

[0565] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from the pre-processed talk paper.

[0566] The frequency of keyword occurrences is calculated, and this data is stored in a database.

[0567] 5. Data Comparison and Analysis

[0568] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews.

[0569] As a concrete example, we use TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[0570] Based on these analysis results, the gap between high-expert crews and low-expert crews will be quantified and evaluated.

[0571] 6. Visualization of Results

[0572] The server visualizes the quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a format that is easy for users to understand.

[0573] This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[0574] 7. Planning and management of training programs

[0575] Based on the results displayed by the system, users develop training plans for low-level crew members.

[0576] The server stores this training plan in a database and provides reminder and notification functions for progress management.

[0577] This allows for continuous tracking of progress towards achieving the training plan and enables adjustments to the plan as needed.

[0578] Specific example

[0579] Example 1:

[0580] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[0581] The server saves the talk paper to the database and performs preprocessing.

[0582] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[0583] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[0584] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[0585] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[0586] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[0587] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[0588] The following describes the processing flow.

[0589] Step 1:

[0590] The user accesses the system, enters their account information, and logs in.

[0591] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[0592] Step 2:

[0593] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[0594] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0595] Step 3:

[0596] The server saves the received talk papers to the database.

[0597] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[0598] Step 4:

[0599] The server passes the saved talk paper to the text analysis program.

[0600] The text analysis program performs preprocessing such as tokenization, stop word removal, and special character deletion.

[0601] Step 5:

[0602] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[0603] The frequency of occurrence of extracted keywords and phrases is calculated and stored in a database.

[0604] Step 6:

[0605] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0606] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[0607] Step 7:

[0608] The server visualizes the data in visual formats such as bar graphs and radar charts, based on the quantified deviation data.

[0609] Send the visualized data to the device.

[0610] Step 8:

[0611] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0612] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[0613] Step 9:

[0614] The user enters the training plan they have created into the system and sends it to the server.

[0615] The server saves the training plan to a database and sets up reminder and notification functions for progress management.

[0616] Step 10:

[0617] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[0618] (Example 1)

[0619] Next, we will describe Example 1. 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."

[0620] Conventional talk paper analysis systems have made it difficult to accurately grasp the performance differences between high-expert and low-level crews and to develop concrete training plans. Furthermore, there was a lack of systems to provide analysis results in a visually easy-to-understand format and to manage the progress of training plans based on those results. Therefore, the inability to achieve efficient and effective crew training remains a challenge.

[0621] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[0622] In this invention, the server includes means for a user to log in and input data on high-expert crew and low-crew; means for a terminal to acquire the input data and send it to the server; means for the server to store the received data in a database and record metadata; means for the server to perform preprocessing on the stored data, such as tokenization and stop word removal; means for the server to extract keywords and important phrases from the preprocessed data; means for the server to compare the frequency of occurrence and usage of the extracted keywords and phrases and quantify the discrepancies; means for the server to visualize the quantified discrepancy data, create graphs and charts and send them to the terminal; means for the user to formulate a training plan based on the visualized results and save it to the server; and means for the server to manage the progress of the training plan and provide reminder and notification functions. This makes it possible to compare high-expert crew and low-crew, and to formulate and manage specific training plans based on the results.

[0623] A "user" refers to a person who accesses the system, enters data, or checks the results.

[0624] "Terminal" refers to a device used by a user, such as a computer or smartphone.

[0625] A "server" refers to a central processing unit that stores, processes, and analyzes data.

[0626] A "database" refers to a system for systematically storing and managing data.

[0627] "Metadata" refers to additional information accompanying the main data, such as the data type and creation date and time.

[0628] "Preprocessing" refers to a series of preparatory tasks performed before data analysis, such as tokenization and stop word removal.

[0629] "Tokenization" refers to the process of dividing text into smaller units such as words and phrases.

[0630] A "stop word" refers to a word that is frequently used in text analysis but does not have any particular meaning.

[0631] A "keyword" refers to a word or short phrase that has significant meaning within a text.

[0632] A "phrase" is a part of language, referring to a combination of multiple words that have a specific meaning on their own.

[0633] "Frequency" refers to an indicator of how often a particular word or phrase is used within a text.

[0634] "Discrepancy" refers to the differences in performance and terminology used between high-expert crews and low-level crews.

[0635] "Quantification" refers to representing data or results using specific numerical values.

[0636] "Visualization" refers to the visual representation of data using graphs and charts.

[0637] A "development plan" refers to a plan designed to improve the skills and performance of low-level crew members.

[0638] A "reminder" refers to an alert that is sent at a specific date or time.

[0639] "Notification function" refers to a system function that transmits information to the user.

[0640] Modes for carrying out the invention

[0641] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[0642] System Configuration

[0643] This system mainly consists of the following elements:

[0644] 1. User terminal

[0645] This is a device used by users to log in and upload talk papers. This includes personal computers, tablets, smartphones, etc.

[0646] 2. Server

[0647] It is a central processing unit that handles data storage, preprocessing, analysis, visualization, and management of development plans.

[0648] 3. Database

[0649] This is a storage device for saving talk papers, analysis results, and training plans. MongoDB will be used here.

[0650] Software to use

[0651] The software configuration of this system is mainly as follows:

[0652] 1. Text Analysis Library

[0653] NLTK (Natural Language Toolkit): Used for tokenizing text and removing stop words.

[0654] 2. Natural Language Processing (NLP) Libraries

[0655] SpaCy: Used to extract keywords and phrases from talk papers.

[0656] 3. Data Visualization Library

[0657] Matplotlib and Plotly are used to visualize quantified deviation data.

[0658] System operation

[0659] 1. Data entry

[0660] The user logs in and uploads the talk papers for the high-expert crew and the low-crew.

[0661] The terminal retrieves the talk paper and sends it to the server.

[0662] 2. Data Storage

[0663] The server saves the received talk papers to a database and records metadata (such as crew type and date / time).

[0664] 3. Text preprocessing

[0665] The server tokenizes the talk paper using NLTK and removes stop words.

[0666] 4. Extraction of keywords and phrases

[0667] The server uses SpaCy to extract keywords and important phrases from the pre-processed talk paper. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[0668] 5. Data Comparison and Analysis

[0669] The server compares the frequency of keywords and phrases used by high-expert crews and low-expert crews.

[0670] The server uses TF-IDF to calculate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[0671] 6. Visualization of Results

[0672] The server visualizes the quantified deviation data using Matplotlib or Plotly to create bar graphs and radar charts.

[0673] The device displays the visualized results to the user.

[0674] 7. Planning and management of training programs

[0675] Based on the results visualized by the user, a training plan for the low-level crew is developed and saved on the server.

[0676] The server manages the progress of the training plan and provides reminders and notifications.

[0677] Specific example

[0678] Example 1:

[0679] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[0680] The server saves the talk paper to the database and performs preprocessing.

[0681] The server uses NLP (Neuro-Linguistic Programming) techniques to extract keywords from the talk papers of Crew A and Crew B.

[0682] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[0683] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[0684] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[0685] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[0686] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[0687] Example of a prompt

[0688] "Analyze the following talk paper and quantify the performance gap between the high-expert crew and the low-expert crew."

[0689] The flow of the specific processing in Example 1 will be explained using Figure 11.

[0690] Step 1:

[0691] The user logs into the system and uploads their talk papers for both the high-expert crew and the low-crew. The inputs here are the user's login information and the talk paper files. The terminal receives this information and sends it to the server. The login information is verified, and if successful, the talk paper upload screen appears. The user clicks the upload button, selects the file, and submits it.

[0692] Step 2:

[0693] The server saves the received talk paper to a database and records metadata (e.g., crew type, upload date and time). The input for this step is the talk paper file and metadata sent from the terminal. The server temporarily stores the file and converts it to a format suitable for storage in the database. It then saves it to the database along with the metadata. The output is the talk paper and metadata stored in the database.

[0694] Step 3:

[0695] The server preprocesses the stored talk papers using the NLTK library. The input for this step is the talk papers loaded from the database. Specifically, the server tokenizes the text (divides it into words) and removes stop words (e.g., "and," "the," etc.). This process formats the talk papers in a way that is suitable for analysis. The output is the preprocessed text data.

[0696] Step 4:

[0697] The server extracts keywords and important phrases from pre-processed text data using the SpaCy library. The input for this step is the pre-processed text data. The server identifies important phrases such as noun phrases and verb phrases and calculates their frequency. Specifically, the server calculates the frequency of occurrence of each keyword and stores it in the database. The output is data of the extracted keywords and their frequencies.

[0698] Step 5:

[0699] The server compares the frequency of keywords and phrases between high-expert and low-expert crews. The input for this step is keyword and phrase frequency data stored in a database. The server calculates TF-IDF to assess the importance of each keyword. It also uses cosine similarity to measure the similarity between talk papers. Specifically, the server quantifies the comparison results and stores them in a database. The output is quantified deviation data.

[0700] Step 6:

[0701] The server visualizes the quantified deviation data using Matplotlib or Plotly. The input for this step is the quantified deviation data. The server creates graphs in visually easy-to-understand formats such as bar graphs and radar charts. Specifically, the server generates graph data and sends it to the terminal. The output is the visualized graph data.

[0702] Step 7:

[0703] The terminal displays the visualized results to the user. The input for this step is the visualized graph data sent from the server. The terminal displays the graph on the screen so that the user can view it. Specifically, the graph is displayed in the terminal's web browser or similar application. The output is the graph displayed to the user.

[0704] Step 8:

[0705] The user develops a training plan for low-level crew members based on the visualized results. The input for this step is visualized graph data. The user fills in a specific training plan in an input form and sends it to the server. Specifically, the user decides on a training plan and enters it into the input form. The output is the training plan data sent to the server.

[0706] Step 9:

[0707] The server stores the training plan in a database and provides reminder and notification functions for progress management. The input for this step is the training plan data submitted by the user. The server stores the training plan in the database and sets reminders and notifications based on the schedule. Specifically, the server sets the notification schedule and sends reminders to the user. The output is the reminders and notifications sent to the user.

[0708] (Application Example 1)

[0709] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0710] Traditionally, the efficiency improvements and training plans for robot operations within factories were primarily relied upon human experience and intuition. This resulted in insufficient transfer of knowledge and skills from experienced operators to new recruits, leading to decreased productivity and increased operational errors. Furthermore, the lack of concrete methods for incorporating the results of talk paper analysis into robot implementation made overall optimization difficult.

[0711] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[0712] In this invention, the server includes means for inputting talk papers of high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to quantify the discrepancies between talk papers and efficiently plan and manage training programs. Furthermore, by linking it with a robot that optimizes operations, it is possible to immediately reflect the analysis results and improve the efficiency of the entire factory.

[0713] A "high-expert crew" is a group of operators who possess advanced expertise and skills in specific tasks or operations.

[0714] A "low crew" is a group of operators who are still in training or have low skill levels in a particular task or job.

[0715] A "talk paper" is a set of procedures, manuals, or customer service guidelines that operators refer to during their work.

[0716] "Input method" refers to the specific hardware or software interface used to input data into a system.

[0717] "Means of preservation" refer to storage devices and database systems for retaining data over the long term.

[0718] "Means of preprocessing" refers to the process of shaping and processing data to facilitate data analysis, and the program that executes it.

[0719] "Means for extracting keywords and phrases" refer to algorithms and techniques for identifying important words and phrases from text data.

[0720] "Methods for quantifying discrepancies" refer to methods and algorithms for measuring the differences between different datasets and expressing them numerically.

[0721] "Means of visualization and displaying results" refers to software or tools for displaying analyzed data in a visual form, such as graphs or charts.

[0722] "Means for formulating and managing training plans" refers to systems and methodologies for developing, tracking, and managing plans for improving operators' skills.

[0723] "Means for linking with a robot and displaying analysis results" refers to an interface and technology for integrating analyzed data into the control system of an industrial robot and displaying the results in real time.

[0724] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system is intended to optimize the operation of robots operating in a factory and is implemented in the following steps.

[0725] System Configuration

[0726] Data entry and saving:

[0727] The user first uploads the talk papers for the high-expert and low-expert crews to the system. The terminal retrieves these talk papers and sends them to the server. The server saves the received talk papers to its database. During this saving process, metadata such as the crew type and date and time are also recorded.

[0728] Text preprocessing and parsing:

[0729] The server first performs preprocessing on the talk papers stored in the database before analyzing them. This preprocessing includes tokenization and stop word removal. Through this process, the talk papers are formatted to be easily analyzed. Furthermore, natural language processing (NLP) techniques are used to extract keywords and important phrases from the preprocessed talk papers. Specific examples include using TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate keyword importance and using cosine similarity to measure the similarity between talk papers.

[0730] Data comparison and analysis:

[0731] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews. Based on this analysis, the discrepancy between high-expert and low-expert crews is quantified and evaluated. The quantified data is stored in a database.

[0732] Visualization and display:

[0733] The server visualizes quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a user-friendly format. This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[0734] Development and management of training plans:

[0735] Based on the results displayed by the system, users create training plans for their low-level crew members. The server stores these training plans in a database and provides reminders and notifications for progress management. This allows for continuous tracking of the training plan's progress and adjustments to the plan as needed.

[0736] Robot collaboration and optimization:

[0737] The server transmits the analysis results to the industrial robots. This allows for real-time display of the analysis results and optimization of operations. This integration supports the overall efficiency of the factory.

[0738] Specific example

[0739] For example, if a high-expert crew uses a talk paper that emphasizes "building trust for customer satisfaction and quick response," and a low-crew uses a talk paper that emphasizes "building trust with customers for quick response," the server compares these talk papers. First, it extracts keywords from each talk paper and calculates the TF-IDF value. Then, it calculates cosine similarity to quantify the divergence between the two talk papers. The quantified data is visualized in bar graphs or radar charts and presented to the user.

[0740] Examples of prompts for generative AI models

[0741] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[0742] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[0743] Step 1:

[0744] The user logs into the system and uploads talk papers for both the high-expert crew and the low-crew. The terminal retrieves these talk papers and sends them to the server. The talk papers are expected to be in file formats such as text files or PDFs.

[0745] Input: Talk papers for the high-expert crew and the low-crew crew (text or PDF format).

[0746] Output: Talk paper sent to the server.

[0747] Step 2:

[0748] The server saves uploaded talk papers to the database. Along with the talk papers, metadata (such as crew type and upload date / time) is also saved. This metadata will be useful for future data analysis.

[0749] Input: Talk paper and metadata sent to the server.

[0750] Output: Talk papers and metadata stored in the database.

[0751] Step 3:

[0752] The server retrieves the talk papers stored in the database for preprocessing. Preprocessing includes tokenization, stop word removal, and normalization. These preprocessing steps prepare the text data for easier analysis.

[0753] Input: Talk paper retrieved from the database.

[0754] Output: Preprocessed text data.

[0755] Step 4:

[0756] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from pre-processed talk papers. Specifically, TF-IDF (Term Frequency-Inverse Document Frequency) is used.

[0757] Input: Preprocessed text data.

[0758] Output: Extracted keywords and phrases, and their frequency data.

[0759] Step 5:

[0760] The server uses TF-IDF to evaluate the importance of each keyword and then uses cosine similarity to measure the similarity between talk papers from high-expert and low-expert crews. This analysis quantifies the discrepancies between talk papers.

[0761] Input: Keywords and phrases, and their frequency data.

[0762] Output: Numerical deviation data.

[0763] Step 6:

[0764] The server visualizes quantified deviation data. It creates bar graphs and radar charts, displaying the results in a user-friendly format. This allows users to visually identify areas that require improvement.

[0765] Input: Numerical deviation data.

[0766] Output: Results visualized as bar graphs or radar charts.

[0767] Step 7:

[0768] The user develops a training plan for the low-level crew based on the results displayed within the system. The server stores this training plan in a database and provides reminders and notifications for progress management.

[0769] Input: Visualized result.

[0770] Output: Training plan stored in the database.

[0771] Step 8:

[0772] The server links the analysis results to the control system of the industrial robot. This linkage allows the analysis results to be reflected in the robot's operation in real time, optimizing the work process.

[0773] Input: Analysis results.

[0774] Output: Analysis results reflected in the robot.

[0775] Specific examples of operation

[0776] If a senior operator's talk paper states "We prioritize trust for customer satisfaction and prompt response," and a junior operator's talk paper states "We build trust with customers to respond quickly," the server analyzes these talk papers. It extracts keywords, compares their frequencies, and derives several characteristics. Then, it quantifies the discrepancies, visualizes them, and provides them to the user. Based on these results, the user creates a training plan, which the server saves to a database. Finally, the analysis results are linked to a robot to optimize operations.

[0777] Examples of prompts for generative AI models

[0778] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[0779] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[0780] This invention combines a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies with an emotion engine. This system not only analyzes the talk papers entered by users and quantifies the discrepancies, but also recognizes and analyzes the user's emotions and reflects them in the training plan, thereby supporting more effective training.

[0781] Program Processing Overview

[0782] 1. Data entry

[0783] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[0784] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0785] 2. Data Storage

[0786] The server saves the received talk papers to the database.

[0787] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[0788] 3. Text preprocessing

[0789] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[0790] 4. Extraction of keywords and phrases

[0791] The server uses natural language processing (NLP) techniques to extract important keywords and phrases from the pre-processed talk paper.

[0792] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[0793] 5. Data Comparison and Analysis

[0794] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0795] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[0796] 6. Application of the Emotion Engine

[0797] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[0798] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[0799] 7. Visualization of Results

[0800] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[0801] Send the visualized data to the device.

[0802] 8. Planning and management of training programs

[0803] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[0804] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[0805] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[0806] Specific example

[0807] Example 1:

[0808] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[0809] The server saves the talk paper to the database and performs preprocessing.

[0810] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[0811] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[0812] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" frequently appear in Crew B's talk paper, the user's emotion will be classified as "negative."

[0813] The server visualizes quantified data and sentiment data in bar graphs and radar charts, and also displays sentiment-related indicators.

[0814] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0815] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[0816] The server automatically adjusts the training plan based on emotional data, adding elements such as mental training and counseling.

[0817] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[0818] This makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, and to create concrete training plans that take into account user sentiment.

[0819] The following describes the processing flow.

[0820] Step 1:

[0821] The user accesses the system, enters their account information, and logs in.

[0822] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[0823] Step 2:

[0824] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[0825] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0826] Step 3:

[0827] The server saves the received talk papers to the database.

[0828] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[0829] Step 4:

[0830] The server passes the saved talk paper to the text analysis program.

[0831] The text analysis program performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[0832] Step 5:

[0833] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[0834] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[0835] Step 6:

[0836] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0837] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[0838] Step 7:

[0839] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[0840] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[0841] Step 8:

[0842] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[0843] Send the visualized data to the device.

[0844] Step 9:

[0845] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0846] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[0847] Step 10:

[0848] The user enters the training plan they have created into the system and sends it to the server.

[0849] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[0850] Step 11:

[0851] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[0852] Step 12:

[0853] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[0854] (Example 2)

[0855] Next, we will describe Example 2. 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".

[0856] Conventional training planning systems made it difficult to compare the performance of high-expert crew members with that of low-level crew members, and they were unable to create specific training plans that took into account user emotions. Furthermore, there was a lack of means to quantify discrepancies in text documents and analyze and visualize emotional data, which hindered effective training support.

[0857] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[0858] In this invention, the server includes means for inputting text documents of high-expert crew and low-crew; means for saving and pre-processing the input text documents; means for extracting keywords and phrases from the pre-processed text documents; means for comparing the frequency and context of the extracted keywords and phrases and quantifying the discrepancies; means for analyzing sentiment from the keywords and phrases and saving the results; means for visualizing the quantified discrepancy data and sentiment data and displaying the results; and means for formulating and managing training plans based on the displayed results. This makes it possible to objectively evaluate the performance differences between high-expert crew and low-crew, formulate specific training plans that take into account user sentiment, and provide effective training support.

[0859] A "text document" refers to a document file saved in text format, including, for example, PDF, Word, and Excel files.

[0860] "Input method" refers to the interface that allows users to upload text documents to the system, and it uses specific software and hardware.

[0861] "Means of saving and pre-processing" refers to the process by which the server saves the received text document to a database and then performs text processing such as tokenization, stop word removal, and removal of special characters.

[0862] "Means for extracting keywords and phrases" refers to the process of identifying and extracting important words and phrases from a text document using natural language processing algorithms.

[0863] "Methods for comparing frequency of occurrence and context and quantifying the discrepancies" refers to the process of analyzing the frequency of occurrence and contextual information of extracted keywords and phrases, and expressing the differences between high-expert crews and low-level crews numerically.

[0864] "Methods for analyzing emotions and saving results" refers to the process of using an emotion engine to identify a user's emotions from keywords and phrases within a text document and recording the results in a database.

[0865] "Means of visualization and displaying results" refers to an interface that visually represents quantified deviation data and sentiment data in the form of graphs and charts, and displays them to the user.

[0866] "Means for formulating and managing training plans" refers to the function of formulating training policies for low-level employees based on visualized data, and tracking and managing their progress.

[0867] This invention combines a system that analyzes text documents (hereinafter referred to as "talk papers") of high-expert crews and low-level crews and quantifies the discrepancies between them with an emotion engine. This system aims to support more effective training by not only analyzing the talk papers entered by users and quantifying the discrepancies, but also recognizing and analyzing the user's emotions and reflecting them in the training plan.

[0868] Hardware and software to be used

[0869] The system consists of the following main hardware and software components.

[0870] Device: A device such as a PC, smartphone, or tablet used by the user to upload their talk paper. Web applications or mobile applications are used.

[0871] Server: A central processing unit for data storage, preprocessing, analysis, and visualization. Server software used includes Apache and Nginx, among others.

[0872] Database: A relational database management system (RDBMS) used to store talk papers and analysis results. MySQL and PostgreSQL are commonly used.

[0873] NLP libraries: Software libraries for natural language processing. These include NLTK, SpaCy, BERT, Word2Vec, and others.

[0874] Emotion engine: APIs or software used for sentiment analysis. Examples include Google Cloud Natural Language API and IBM Watson.

[0875] Graph libraries: Tools for visualizing data. D3.js and Chart.js are commonly used.

[0876] Specific processing flow

[0877] 1. Data entry

[0878] Users log into the system and upload the talk papers for their high-expert and low-expert crews. For example, a user selects the talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf) from their PC or smartphone and clicks the upload button.

[0879] 2. Data storage and preprocessing

[0880] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[0881] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[0882] 3. Keyword and phrase extraction

[0883] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from pre-processed talk papers. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[0884] 4. Data Comparison and Analysis

[0885] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[0886] The server uses the TF-IDF algorithm to evaluate the importance of keywords, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[0887] 5. Application of the Emotion Engine

[0888] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" appear frequently, the user's emotion will be classified as "negative."

[0889] 6. Visualization of Results

[0890] The server visualizes quantified deviation and sentiment data. In addition to bar graphs and radar charts, it displays sentiment-related indicators.

[0891] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[0892] 7. Planning and management of training programs

[0893] Based on the results visualized by the user, a training plan for low-level crew members will be developed. For example, "mental training courses" and "communication skills improvement sessions" will be considered.

[0894] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[0895] A possible example of a specific prompt might be, "Extract keywords from the talk papers of the high-expert crew and low-expert crew, analyze their emotions, and incorporate them into the training plan."

[0896] This series of processes makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, develop specific training plans that take user emotions into consideration, and provide effective training support.

[0897] The flow of the specific processing in Example 2 will be explained using Figure 13.

[0898] Step 1:

[0899] Data entry

[0900] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[0901] Input: Talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf)

[0902] The device retrieves the uploaded talk paper as text data and sends it to the server.

[0903] Output: Text data sent to the server

[0904] Specific actions:

[0905] The user selects a talk paper from their PC or smartphone and clicks the upload button. The device reads the file and sends it to the server as text data.

[0906] Step 2:

[0907] Data storage and preprocessing

[0908] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[0909] Input: Text data sent to the server

[0910] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[0911] Output: Preprocessed text data

[0912] Specific actions:

[0913] The server saves the talk paper to the database, recording metadata such as the file name, upload date and time, and user ID. Next, the server uses a text analysis program to perform tokenization, stop word removal, and special character removal to generate pre-processed text data.

[0914] Step 3:

[0915] Keyword and phrase extraction

[0916] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from the pre-processed talk paper.

[0917] Input: Preprocessed text data

[0918] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[0919] Output: List of keywords and phrases and their frequencies

[0920] Specific actions:

[0921] The server applies natural language processing technology to extract important keywords (e.g., "customer satisfaction," "prompt response") and phrases from the talk paper. It calculates the frequency of occurrence of the extracted keywords and stores the results in a database.

[0922] Step 4:

[0923] Data comparison and analysis

[0924] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[0925] Input: A list of keywords and phrases and their frequency of occurrence.

[0926] The server uses cosine similarity to calculate the similarity between the two talk papers and quantifies the discrepancy.

[0927] Output: Quantified deviation data

[0928] Specific actions:

[0929] The server applies the TF-IDF algorithm to calculate the importance of keywords within the talk paper. Next, it calculates cosine similarity to quantify the discrepancy between the talk papers of the high-expert crew and the low-crew crew.

[0930] Step 5:

[0931] Application of the emotion engine

[0932] The server uses an emotion engine (e.g., Google Cloud Natural Language API or IBM Watson) to analyze the user's emotions from keywords and phrases within the talk paper.

[0933] Input: Preprocessed text data

[0934] The server classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[0935] Output: Sentiment analysis result data

[0936] Specific actions:

[0937] The server uses an emotion engine to analyze keywords and phrases within the talk paper and determine an emotion score and emotion type. The results are then stored in a database.

[0938] Step 6:

[0939] Visualization of results

[0940] The server visualizes quantified deviation data and sentiment data.

[0941] Input: Quantified deviation data and sentiment analysis result data

[0942] The server uses a graphing library (e.g., D3.js or Chart.js) to visualize this data in bar graphs or radar charts.

[0943] Output: Visualized graphs and charts

[0944] Specific actions:

[0945] The server generates graphs based on deviation data and sentiment data, prepares data for display in a visually easy-to-understand format, and sends this data to the terminal.

[0946] Step 7:

[0947] Development and management of training plans

[0948] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[0949] Input: Visualized graphs and charts

[0950] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[0951] Output: Training plan and progress management data

[0952] Specific actions:

[0953] The user views the displayed graph data, creates a specific training plan for the low-level crew member, and saves it to the server. The server then takes sentiment data into consideration and automatically adjusts the training plan, setting up relevant reminders and notifications.

[0954] (Application Example 2)

[0955] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server," and the smart glasses 214 will be referred to as the "terminal."

[0956] Previous methods for evaluating staff performance and developing training plans relied heavily on subjective assessments, making effective training difficult. Furthermore, failing to consider staff emotional states during training planning could lead to decreased motivation and stress. This invention aims to provide more effective training plans by quantifying the discrepancies between high-expert and low-level staff communication reports and by considering staff emotional states.

[0957] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[0958] In this invention, the server includes means for inputting talk papers from high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to visualize quantified discrepancy data and sentiment data, formulate and manage training plans based on the results, and periodically monitor and notify progress. Furthermore, by including means for analyzing the user's emotions from keywords and phrases in the talk papers using an emotion analysis engine, it becomes possible to formulate training plans that reflect the emotional state of the staff.

[0959] A "high-expert crew" is a staff member who possesses high levels of experience and skills and demonstrates outstanding performance in their work.

[0960] A "low-level crew" refers to staff members who have relatively low experience and skills, and who still have room for improvement in their work.

[0961] A "talk paper" refers to a document or script that staff members use when responding to customers or giving explanations.

[0962] "Means of input" refers to devices or interfaces used to import staff members' talk papers into the system.

[0963] "Means of storage" refers to storage or databases used to record and retain entered data.

[0964] "Methods for preprocessing" refer to processes such as text tokenization, stop word removal, and special character deletion performed to prepare the input data into a format that is easy to analyze.

[0965] "Means for extracting keywords and phrases" refers to methods for extracting important words and phrases using natural language processing algorithms.

[0966] "Methods for quantifying discrepancies" refer to algorithms and analytical methods for quantitatively evaluating and expressing the differences between talk papers from high-expert crews and low-expert crews as numerical values.

[0967] "Means of visualization" refer to tools and software used to display analysis results in graphical formats such as bar graphs and radar charts.

[0968] "Means for formulating and managing training plans" refers to systems and mechanisms for formulating staff training plans, monitoring their progress, and issuing necessary notifications and reminders.

[0969] An "emotion analysis engine" refers to algorithms and technologies that read emotions from text data and classify them as positive, negative, neutral, etc.

[0970] "Monitoring and notification mechanisms" refer to functions that allow the system to periodically check the progress of the training plan and generate reminders and alerts as needed.

[0971] This invention consists of a system for inputting talk papers from high-expert crew members and low-level crew members, quantifying the discrepancies between them, and formulating and managing staff training plans. In addition, by using an emotion analysis engine to analyze the user's emotions and reflecting them in the training plan, it is possible to support more effective training.

[0972] Hardware and software to be used

[0973] Hardware:

[0974] Server: Used for processing and storing data.

[0975] Device: An input device such as a smartphone or smart glasses.

[0976] software:

[0977] Speech recognition software (Google Speech Recognition): Converts spoken conversation into text data.

[0978] Natural language processing libraries (nltk, scikit-learn): Used for text tokenization, stop word removal, special character removal, and keyword extraction.

[0979] Database: Used to store talk papers and analysis results.

[0980] Emotion analysis engine: Recognizes and analyzes emotions such as positive, negative, and neutral.

[0981] Program Processing Overview

[0982] Data entry and saving

[0983] Users upload talk papers for high-expert and low-level crews to the system using their smartphones or smart glasses. The server receives this data and stores it in a database. At the same time, it also records metadata (crew type, date and time, etc.).

[0984] Text preprocessing

[0985] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and special character deletion. This prepares the text for easier analysis.

[0986] Keyword and phrase extraction and comparison

[0987] The server uses natural language processing techniques to extract important keywords and phrases from pre-processed talk papers. Then, it evaluates the importance of the keywords using the TF-IDF algorithm, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[0988] Emotion analysis

[0989] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases in the talk paper. It classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[0990] Visualization of results

[0991] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts. This visualized data is sent to the terminal, and the results are displayed to the user.

[0992] Development and management of training plans

[0993] Based on the user's visualized results, the server develops a training plan for the low-crew member. The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it incorporates elements of mental support. The server supports the progress management of the training plan and provides necessary reminders and notifications.

[0994] Specific examples of implementation

[0995] For example, imagine a system where smart glasses are used to record conversations between staff working in a physical store and customers, and the data is analyzed. The recorded conversations are sent to a server and compared to the talk papers of high-expert crew members. Based on the analysis results, a staff training plan is developed. The progress of the plan is then monitored regularly, and notifications are sent to the user as needed. If the emotional data is negative, mental training is suggested.

[0996] Example of a prompt

[0997] "Please analyze the conversation text, calculate its similarity to advanced customer service skills, and perform sentiment analysis. The following conversation text: {Conversation Text}"

[0998] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[0999] Step 1:

[1000] Data entry

[1001] Users record the talk papers of high-expert and low-crew members using their smartphones or smart glasses and upload them to the system. The device converts the recorded audio data into text data and sends it to the server.

[1002] Input: Audio data

[1003] Output: Text data

[1004] Specific operation: Use speech recognition software (Google Speech Recognition) to convert speech data into text.

[1005] Step 2:

[1006] Data storage

[1007] The server stores the received text data in a database. At the same time, it also records metadata related to the talk paper (such as crew type and date / time).

[1008] Input: Text data, metadata

[1009] Output: Saved database entries

[1010] Specific operation: Connect to the database and save text data and metadata in a specified format.

[1011] Step 3:

[1012] Pre-treatment

[1013] The server retrieves talk papers from the database and performs preprocessing. Specifically, it performs tokenization, stop word removal, and removal of special characters.

[1014] Input: Text data

[1015] Output: Preprocessed text data

[1016] Specific operation: Uses the natural language processing library (nltk) to perform tokenization, stop word removal, and special character removal.

[1017] Step 4:

[1018] Keyword and phrase extraction

[1019] The server extracts important keywords and phrases from the pre-processed text data.

[1020] Input: Preprocessed text data

[1021] Output: List of keywords and phrases

[1022] Specific operation: Use TfidfVectorizer to evaluate and extract keywords and phrases from text based on their importance.

[1023] Step 5:

[1024] Quantifying the discrepancy

[1025] The server uses the TF-IDF algorithm and cosine similarity to calculate the similarity between the talk papers of the high-expert crew and the low-crew, and quantifies the discrepancy.

[1026] Input: List of keywords and phrases

[1027] Output: Quantified deviation data

[1028] Specific operation: Calculates similarity using TfidfVectorizer and cosine_similarity.

[1029] Step 6:

[1030] Emotion analysis

[1031] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases within the talk paper.

[1032] Input: List of keywords and phrases

[1033] Output: Sentiment analysis results (positive, negative, neutral, etc.)

[1034] Specific operation: Classifies the sentiment of keywords and phrases using a proprietary sentiment analysis algorithm.

[1035] Step 7:

[1036] Visualization of results

[1037] The server visualizes quantified deviation data and sentiment data and sends the results to the terminal.

[1038] Input: Quantified deviation data, sentiment data

[1039] Output: Visualized data (graphs, charts, etc.)

[1040] Specific operation: Use matplotlib to generate and visualize bar graphs and radar charts.

[1041] Step 8:

[1042] Development and management of training plans

[1043] Users create training plans for their low-level crew members based on visualized results. The server automatically adjusts the training plan based on sentiment data, monitors progress, and provides notifications.

[1044] Input: Visualized data, training plan

[1045] Output: Adjusted training plan, progress notifications

[1046] Specific operation: The server adjusts the care plan based on sentiment data stored in the database, monitors progress, and generates reminders and notifications as needed.

[1047] 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 user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1048] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1049] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the smart glasses 214.

[1050] [Third Embodiment]

[1051] Figure 5 shows an example of the configuration of the data processing system 310 according to the third embodiment.

[1052] As shown in Figure 5, the data processing system 310 includes a data processing device 12 and a headset terminal 314. An example of the data processing device 12 is a server.

[1053] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1054] The headset terminal 314 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a display 343. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and display 343 are also connected to the bus 52.

[1055] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1056] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1057] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1058] Figure 6 shows an example of the main functions of the data processing device 12 and the headset terminal 314. As shown in Figure 6, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1059] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[1060] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1061] In the headset terminal 314, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1062] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the headset terminal 314 will be referred to as the "terminal".

[1063] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[1064] Program Processing Overview

[1065] 1. Data entry

[1066] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[1067] The device retrieves the uploaded talk paper and sends it to the server.

[1068] 2. Data Storage

[1069] The server saves the received talk papers to the database.

[1070] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[1071] 3. Text preprocessing

[1072] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization and stop word removal.

[1073] This formats the talk paper in a way that makes it easier to analyze.

[1074] 4. Extraction of keywords and phrases

[1075] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from the pre-processed talk paper.

[1076] The frequency of keyword occurrences is calculated, and this data is stored in a database.

[1077] 5. Data Comparison and Analysis

[1078] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews.

[1079] As a concrete example, we use TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[1080] Based on these analysis results, the gap between high-expert crews and low-expert crews will be quantified and evaluated.

[1081] 6. Visualization of Results

[1082] The server visualizes the quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a format that is easy for users to understand.

[1083] This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[1084] 7. Planning and management of training programs

[1085] Based on the results displayed by the system, users develop training plans for low-level crew members.

[1086] The server stores this training plan in a database and provides reminder and notification functions for progress management.

[1087] This allows for continuous tracking of progress towards achieving the training plan and enables adjustments to the plan as needed.

[1088] Specific example

[1089] Example 1:

[1090] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[1091] The server saves the talk paper to the database and performs preprocessing.

[1092] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[1093] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[1094] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[1095] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[1096] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[1097] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[1098] The following describes the processing flow.

[1099] Step 1:

[1100] The user accesses the system, enters their account information, and logs in.

[1101] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[1102] Step 2:

[1103] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[1104] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1105] Step 3:

[1106] The server saves the received talk papers to the database.

[1107] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[1108] Step 4:

[1109] The server passes the saved talk paper to the text analysis program.

[1110] The text analysis program performs preprocessing such as tokenization, stop word removal, and special character deletion.

[1111] Step 5:

[1112] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[1113] The frequency of occurrence of extracted keywords and phrases is calculated and stored in a database.

[1114] Step 6:

[1115] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1116] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[1117] Step 7:

[1118] The server visualizes the data in visual formats such as bar graphs and radar charts, based on the quantified deviation data.

[1119] Send the visualized data to the device.

[1120] Step 8:

[1121] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1122] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[1123] Step 9:

[1124] The user enters the training plan they have created into the system and sends it to the server.

[1125] The server saves the training plan to a database and sets up reminder and notification functions for progress management.

[1126] Step 10:

[1127] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[1128] (Example 1)

[1129] Next, we will describe Example 1. 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."

[1130] Conventional talk paper analysis systems have made it difficult to accurately grasp the performance differences between high-expert and low-level crews and to develop concrete training plans. Furthermore, there was a lack of systems to provide analysis results in a visually easy-to-understand format and to manage the progress of training plans based on those results. Therefore, the inability to achieve efficient and effective crew training remains a challenge.

[1131] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1132] In this invention, the server includes means for a user to log in and input data on high-expert crew and low-crew; means for a terminal to acquire the input data and send it to the server; means for the server to store the received data in a database and record metadata; means for the server to perform preprocessing on the stored data, such as tokenization and stop word removal; means for the server to extract keywords and important phrases from the preprocessed data; means for the server to compare the frequency of occurrence and usage of the extracted keywords and phrases and quantify the discrepancies; means for the server to visualize the quantified discrepancy data, create graphs and charts and send them to the terminal; means for the user to formulate a training plan based on the visualized results and save it to the server; and means for the server to manage the progress of the training plan and provide reminder and notification functions. This makes it possible to compare high-expert crew and low-crew, and to formulate and manage specific training plans based on the results.

[1133] A "user" refers to a person who accesses the system, enters data, or checks the results.

[1134] "Terminal" refers to a device used by a user, such as a computer or smartphone.

[1135] A "server" refers to a central processing unit that stores, processes, and analyzes data.

[1136] A "database" refers to a system for systematically storing and managing data.

[1137] "Metadata" refers to additional information accompanying the main data, such as the data type and creation date and time.

[1138] "Preprocessing" refers to a series of preparatory tasks performed before data analysis, such as tokenization and stop word removal.

[1139] "Tokenization" refers to the process of dividing text into smaller units such as words and phrases.

[1140] A "stop word" refers to a word that is frequently used in text analysis but does not have any particular meaning.

[1141] A "keyword" refers to a word or short phrase that has significant meaning within a text.

[1142] A "phrase" is a part of language, referring to a combination of multiple words that have a specific meaning on their own.

[1143] "Frequency" refers to an indicator of how often a particular word or phrase is used within a text.

[1144] "Discrepancy" refers to the differences in performance and terminology used between high-expert crews and low-level crews.

[1145] "Quantification" refers to representing data or results using specific numerical values.

[1146] "Visualization" refers to the visual representation of data using graphs and charts.

[1147] A "development plan" refers to a plan designed to improve the skills and performance of low-level crew members.

[1148] A "reminder" refers to an alert that is sent at a specific date or time.

[1149] "Notification function" refers to a system function that transmits information to the user.

[1150] Modes for carrying out the invention

[1151] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[1152] System Configuration

[1153] This system mainly consists of the following elements:

[1154] 1. User terminal

[1155] This is a device used by users to log in and upload talk papers. This includes personal computers, tablets, smartphones, etc.

[1156] 2. Server

[1157] It is a central processing unit that handles data storage, preprocessing, analysis, visualization, and management of development plans.

[1158] 3. Database

[1159] This is a storage device for saving talk papers, analysis results, and training plans. MongoDB will be used here.

[1160] Software to use

[1161] The software configuration of this system is mainly as follows:

[1162] 1. Text Analysis Library

[1163] NLTK (Natural Language Toolkit): Used for tokenizing text and removing stop words.

[1164] 2. Natural Language Processing (NLP) Libraries

[1165] SpaCy: Used to extract keywords and phrases from talk papers.

[1166] 3. Data Visualization Library

[1167] Matplotlib and Plotly are used to visualize quantified deviation data.

[1168] System operation

[1169] 1. Data entry

[1170] The user logs in and uploads the talk papers for the high-expert crew and the low-crew.

[1171] The terminal retrieves the talk paper and sends it to the server.

[1172] 2. Data Storage

[1173] The server saves the received talk papers to a database and records metadata (such as crew type and date / time).

[1174] 3. Text preprocessing

[1175] The server tokenizes the talk paper using NLTK and removes stop words.

[1176] 4. Extraction of keywords and phrases

[1177] The server uses SpaCy to extract keywords and important phrases from the pre-processed talk paper. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[1178] 5. Data Comparison and Analysis

[1179] The server compares the frequency of keywords and phrases used by high-expert crews and low-expert crews.

[1180] The server uses TF-IDF to calculate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[1181] 6. Visualization of Results

[1182] The server visualizes the quantified deviation data using Matplotlib or Plotly to create bar graphs and radar charts.

[1183] The device displays the visualized results to the user.

[1184] 7. Planning and management of training programs

[1185] Based on the results visualized by the user, a training plan for the low-level crew is developed and saved on the server.

[1186] The server manages the progress of the training plan and provides reminders and notifications.

[1187] Specific example

[1188] Example 1:

[1189] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[1190] The server saves the talk paper to the database and performs preprocessing.

[1191] The server uses NLP (Neuro-Linguistic Programming) techniques to extract keywords from the talk papers of Crew A and Crew B.

[1192] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[1193] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[1194] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[1195] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[1196] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[1197] Example of a prompt

[1198] "Analyze the following talk paper and quantify the performance gap between the high-expert crew and the low-expert crew."

[1199] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1200] Step 1:

[1201] The user logs into the system and uploads their talk papers for both the high-expert crew and the low-crew. The inputs here are the user's login information and the talk paper files. The terminal receives this information and sends it to the server. The login information is verified, and if successful, the talk paper upload screen appears. The user clicks the upload button, selects the file, and submits it.

[1202] Step 2:

[1203] The server saves the received talk paper to a database and records metadata (e.g., crew type, upload date and time). The input for this step is the talk paper file and metadata sent from the terminal. The server temporarily stores the file and converts it to a format suitable for storage in the database. It then saves it to the database along with the metadata. The output is the talk paper and metadata stored in the database.

[1204] Step 3:

[1205] The server preprocesses the stored talk papers using the NLTK library. The input for this step is the talk papers loaded from the database. Specifically, the server tokenizes the text (divides it into words) and removes stop words (e.g., "and," "the," etc.). This process formats the talk papers in a way that is suitable for analysis. The output is the preprocessed text data.

[1206] Step 4:

[1207] The server extracts keywords and important phrases from pre-processed text data using the SpaCy library. The input for this step is the pre-processed text data. The server identifies important phrases such as noun phrases and verb phrases and calculates their frequency. Specifically, the server calculates the frequency of occurrence of each keyword and stores it in the database. The output is data of the extracted keywords and their frequencies.

[1208] Step 5:

[1209] The server compares the frequency of keywords and phrases between high-expert and low-expert crews. The input for this step is keyword and phrase frequency data stored in a database. The server calculates TF-IDF to assess the importance of each keyword. It also uses cosine similarity to measure the similarity between talk papers. Specifically, the server quantifies the comparison results and stores them in a database. The output is quantified deviation data.

[1210] Step 6:

[1211] The server visualizes the quantified deviation data using Matplotlib or Plotly. The input for this step is the quantified deviation data. The server creates graphs in visually easy-to-understand formats such as bar graphs and radar charts. Specifically, the server generates graph data and sends it to the terminal. The output is the visualized graph data.

[1212] Step 7:

[1213] The terminal displays the visualized results to the user. The input for this step is the visualized graph data sent from the server. The terminal displays the graph on the screen so that the user can view it. Specifically, the graph is displayed in the terminal's web browser or similar application. The output is the graph displayed to the user.

[1214] Step 8:

[1215] The user develops a training plan for low-level crew members based on the visualized results. The input for this step is visualized graph data. The user fills in a specific training plan in an input form and sends it to the server. Specifically, the user decides on a training plan and enters it into the input form. The output is the training plan data sent to the server.

[1216] Step 9:

[1217] The server stores the training plan in a database and provides reminder and notification functions for progress management. The input for this step is the training plan data submitted by the user. The server stores the training plan in the database and sets reminders and notifications based on the schedule. Specifically, the server sets the notification schedule and sends reminders to the user. The output is the reminders and notifications sent to the user.

[1218] (Application Example 1)

[1219] Next, we will explain Application Example 1. In the following explanation, 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."

[1220] Traditionally, the efficiency improvements and training plans for robot operations within factories were primarily relied upon human experience and intuition. This resulted in insufficient transfer of knowledge and skills from experienced operators to new recruits, leading to decreased productivity and increased operational errors. Furthermore, the lack of concrete methods for incorporating the results of talk paper analysis into robot implementation made overall optimization difficult.

[1221] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1222] In this invention, the server includes means for inputting talk papers of high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to quantify the discrepancies between talk papers and efficiently plan and manage training programs. Furthermore, by linking it with a robot that optimizes operations, it is possible to immediately reflect the analysis results and improve the efficiency of the entire factory.

[1223] A "high-expert crew" is a group of operators who possess advanced expertise and skills in specific tasks or operations.

[1224] A "low crew" is a group of operators who are still in training or have low skill levels in a particular task or job.

[1225] A "talk paper" is a set of procedures, manuals, or customer service guidelines that operators refer to during their work.

[1226] "Input method" refers to the specific hardware or software interface used to input data into a system.

[1227] "Means of preservation" refer to storage devices and database systems for retaining data over the long term.

[1228] "Means of preprocessing" refers to the process of shaping and processing data to facilitate data analysis, and the program that executes it.

[1229] "Means for extracting keywords and phrases" refer to algorithms and techniques for identifying important words and phrases from text data.

[1230] "Methods for quantifying discrepancies" refer to methods and algorithms for measuring the differences between different datasets and expressing them numerically.

[1231] "Means of visualization and displaying results" refers to software or tools for displaying analyzed data in a visual form, such as graphs or charts.

[1232] "Means for formulating and managing training plans" refers to systems and methodologies for developing, tracking, and managing plans for improving operators' skills.

[1233] "Means for linking with a robot and displaying analysis results" refers to an interface and technology for integrating analyzed data into the control system of an industrial robot and displaying the results in real time.

[1234] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system is intended to optimize the operation of robots operating in a factory and is implemented in the following steps.

[1235] System Configuration

[1236] Data entry and saving:

[1237] The user first uploads the talk papers for the high-expert and low-expert crews to the system. The terminal retrieves these talk papers and sends them to the server. The server saves the received talk papers to its database. During this saving process, metadata such as the crew type and date and time are also recorded.

[1238] Text preprocessing and parsing:

[1239] The server first performs preprocessing on the talk papers stored in the database before analyzing them. This preprocessing includes tokenization and stop word removal. Through this process, the talk papers are formatted to be easily analyzed. Furthermore, natural language processing (NLP) techniques are used to extract keywords and important phrases from the preprocessed talk papers. Specific examples include using TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate keyword importance and using cosine similarity to measure the similarity between talk papers.

[1240] Data comparison and analysis:

[1241] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews. Based on this analysis, the discrepancy between high-expert and low-expert crews is quantified and evaluated. The quantified data is stored in a database.

[1242] Visualization and display:

[1243] The server visualizes quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a user-friendly format. This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[1244] Development and management of training plans:

[1245] Based on the results displayed by the system, users create training plans for their low-level crew members. The server stores these training plans in a database and provides reminders and notifications for progress management. This allows for continuous tracking of the training plan's progress and adjustments to the plan as needed.

[1246] Robot collaboration and optimization:

[1247] The server transmits the analysis results to the industrial robots. This allows for real-time display of the analysis results and optimization of operations. This integration supports the overall efficiency of the factory.

[1248] Specific example

[1249] For example, if a high-expert crew uses a talk paper that emphasizes "building trust for customer satisfaction and quick response," and a low-crew uses a talk paper that emphasizes "building trust with customers for quick response," the server compares these talk papers. First, it extracts keywords from each talk paper and calculates the TF-IDF value. Then, it calculates cosine similarity to quantify the divergence between the two talk papers. The quantified data is visualized in bar graphs or radar charts and presented to the user.

[1250] Examples of prompts for generative AI models

[1251] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[1252] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1253] Step 1:

[1254] The user logs into the system and uploads talk papers for both the high-expert crew and the low-crew. The terminal retrieves these talk papers and sends them to the server. The talk papers are expected to be in file formats such as text files or PDFs.

[1255] Input: Talk papers for the high-expert crew and the low-crew crew (text or PDF format).

[1256] Output: Talk paper sent to the server.

[1257] Step 2:

[1258] The server saves uploaded talk papers to the database. Along with the talk papers, metadata (such as crew type and upload date / time) is also saved. This metadata will be useful for future data analysis.

[1259] Input: Talk paper and metadata sent to the server.

[1260] Output: Talk papers and metadata stored in the database.

[1261] Step 3:

[1262] The server retrieves the talk papers stored in the database for preprocessing. Preprocessing includes tokenization, stop word removal, and normalization. These preprocessing steps prepare the text data for easier analysis.

[1263] Input: Talk paper retrieved from the database.

[1264] Output: Preprocessed text data.

[1265] Step 4:

[1266] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from pre-processed talk papers. Specifically, TF-IDF (Term Frequency-Inverse Document Frequency) is used.

[1267] Input: Preprocessed text data.

[1268] Output: Extracted keywords and phrases, and their frequency data.

[1269] Step 5:

[1270] The server uses TF-IDF to evaluate the importance of each keyword and then uses cosine similarity to measure the similarity between talk papers from high-expert and low-expert crews. This analysis quantifies the discrepancies between talk papers.

[1271] Input: Keywords and phrases, and their frequency data.

[1272] Output: Numerical deviation data.

[1273] Step 6:

[1274] The server visualizes quantified deviation data. It creates bar graphs and radar charts, displaying the results in a user-friendly format. This allows users to visually identify areas that require improvement.

[1275] Input: Numerical deviation data.

[1276] Output: Results visualized as bar graphs or radar charts.

[1277] Step 7:

[1278] The user develops a training plan for the low-level crew based on the results displayed within the system. The server stores this training plan in a database and provides reminders and notifications for progress management.

[1279] Input: Visualized result.

[1280] Output: Training plan stored in the database.

[1281] Step 8:

[1282] The server links the analysis results to the control system of the industrial robot. This linkage allows the analysis results to be reflected in the robot's operation in real time, optimizing the work process.

[1283] Input: Analysis results.

[1284] Output: Analysis results reflected in the robot.

[1285] Specific examples of operation

[1286] If a senior operator's talk paper states "We prioritize trust for customer satisfaction and prompt response," and a junior operator's talk paper states "We build trust with customers to respond quickly," the server analyzes these talk papers. It extracts keywords, compares their frequencies, and derives several characteristics. Then, it quantifies the discrepancies, visualizes them, and provides them to the user. Based on these results, the user creates a training plan, which the server saves to a database. Finally, the analysis results are linked to a robot to optimize operations.

[1287] Examples of prompts for generative AI models

[1288] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[1289] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1290] This invention combines a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies with an emotion engine. This system not only analyzes the talk papers entered by users and quantifies the discrepancies, but also recognizes and analyzes the user's emotions and reflects them in the training plan, thereby supporting more effective training.

[1291] Program Processing Overview

[1292] 1. Data entry

[1293] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[1294] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1295] 2. Data Storage

[1296] The server saves the received talk papers to the database.

[1297] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[1298] 3. Text preprocessing

[1299] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[1300] 4. Extraction of keywords and phrases

[1301] The server uses natural language processing (NLP) techniques to extract important keywords and phrases from the pre-processed talk paper.

[1302] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[1303] 5. Data Comparison and Analysis

[1304] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1305] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[1306] 6. Application of the Emotion Engine

[1307] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[1308] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[1309] 7. Visualization of Results

[1310] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[1311] Send the visualized data to the device.

[1312] 8. Planning and management of training programs

[1313] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[1314] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[1315] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[1316] Specific example

[1317] Example 1:

[1318] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[1319] The server saves the talk paper to the database and performs preprocessing.

[1320] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[1321] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[1322] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" frequently appear in Crew B's talk paper, the user's emotion will be classified as "negative."

[1323] The server visualizes quantified data and sentiment data in bar graphs and radar charts, and also displays sentiment-related indicators.

[1324] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1325] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[1326] The server automatically adjusts the training plan based on emotional data, adding elements such as mental training and counseling.

[1327] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[1328] This makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, and to create concrete training plans that take into account user sentiment.

[1329] The following describes the processing flow.

[1330] Step 1:

[1331] The user accesses the system, enters their account information, and logs in.

[1332] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[1333] Step 2:

[1334] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[1335] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1336] Step 3:

[1337] The server saves the received talk papers to the database.

[1338] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[1339] Step 4:

[1340] The server passes the saved talk paper to the text analysis program.

[1341] The text analysis program performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[1342] Step 5:

[1343] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[1344] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[1345] Step 6:

[1346] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1347] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[1348] Step 7:

[1349] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[1350] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[1351] Step 8:

[1352] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[1353] Send the visualized data to the device.

[1354] Step 9:

[1355] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1356] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[1357] Step 10:

[1358] The user enters the training plan they have created into the system and sends it to the server.

[1359] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[1360] Step 11:

[1361] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[1362] Step 12:

[1363] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[1364] (Example 2)

[1365] Next, we will describe Example 2. 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."

[1366] Conventional training planning systems made it difficult to compare the performance of high-expert crew members with that of low-level crew members, and they were unable to create specific training plans that took into account user emotions. Furthermore, there was a lack of means to quantify discrepancies in text documents and analyze and visualize emotional data, which hindered effective training support.

[1367] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1368] In this invention, the server includes means for inputting text documents of high-expert crew and low-crew; means for saving and pre-processing the input text documents; means for extracting keywords and phrases from the pre-processed text documents; means for comparing the frequency and context of the extracted keywords and phrases and quantifying the discrepancies; means for analyzing sentiment from the keywords and phrases and saving the results; means for visualizing the quantified discrepancy data and sentiment data and displaying the results; and means for formulating and managing training plans based on the displayed results. This makes it possible to objectively evaluate the performance differences between high-expert crew and low-crew, formulate specific training plans that take into account user sentiment, and provide effective training support.

[1369] A "text document" refers to a document file saved in text format, including, for example, PDF, Word, and Excel files.

[1370] "Input method" refers to the interface that allows users to upload text documents to the system, and it uses specific software and hardware.

[1371] "Means of saving and pre-processing" refers to the process by which the server saves the received text document to a database and then performs text processing such as tokenization, stop word removal, and removal of special characters.

[1372] "Means for extracting keywords and phrases" refers to the process of identifying and extracting important words and phrases from a text document using natural language processing algorithms.

[1373] "Methods for comparing frequency of occurrence and context and quantifying the discrepancies" refers to the process of analyzing the frequency of occurrence and contextual information of extracted keywords and phrases, and expressing the differences between high-expert crews and low-level crews numerically.

[1374] "Methods for analyzing emotions and saving results" refers to the process of using an emotion engine to identify a user's emotions from keywords and phrases within a text document and recording the results in a database.

[1375] "Means of visualization and displaying results" refers to an interface that visually represents quantified deviation data and sentiment data in the form of graphs and charts, and displays them to the user.

[1376] "Means for formulating and managing training plans" refers to the function of formulating training policies for low-level employees based on visualized data, and tracking and managing their progress.

[1377] This invention combines a system that analyzes text documents (hereinafter referred to as "talk papers") of high-expert crews and low-level crews and quantifies the discrepancies between them with an emotion engine. This system aims to support more effective training by not only analyzing the talk papers entered by users and quantifying the discrepancies, but also recognizing and analyzing the user's emotions and reflecting them in the training plan.

[1378] Hardware and software to be used

[1379] The system consists of the following main hardware and software components.

[1380] Device: A device such as a PC, smartphone, or tablet used by the user to upload their talk paper. Web applications or mobile applications are used.

[1381] Server: A central processing unit for data storage, preprocessing, analysis, and visualization. Server software used includes Apache and Nginx, among others.

[1382] Database: A relational database management system (RDBMS) used to store talk papers and analysis results. MySQL and PostgreSQL are commonly used.

[1383] NLP libraries: Software libraries for natural language processing. These include NLTK, SpaCy, BERT, Word2Vec, and others.

[1384] Emotion engine: APIs or software used for sentiment analysis. Examples include Google Cloud Natural Language API and IBM Watson.

[1385] Graph libraries: Tools for visualizing data. D3.js and Chart.js are commonly used.

[1386] Specific processing flow

[1387] 1. Data entry

[1388] Users log into the system and upload the talk papers for their high-expert and low-expert crews. For example, a user selects the talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf) from their PC or smartphone and clicks the upload button.

[1389] 2. Data storage and preprocessing

[1390] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[1391] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[1392] 3. Keyword and phrase extraction

[1393] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from pre-processed talk papers. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[1394] 4. Data Comparison and Analysis

[1395] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[1396] The server uses the TF-IDF algorithm to evaluate the importance of keywords, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[1397] 5. Application of the Emotion Engine

[1398] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" appear frequently, the user's emotion will be classified as "negative."

[1399] 6. Visualization of Results

[1400] The server visualizes quantified deviation and sentiment data. In addition to bar graphs and radar charts, it displays sentiment-related indicators.

[1401] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1402] 7. Planning and management of training programs

[1403] Based on the results visualized by the user, a training plan for low-level crew members will be developed. For example, "mental training courses" and "communication skills improvement sessions" will be considered.

[1404] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[1405] A possible example of a specific prompt might be, "Extract keywords from the talk papers of the high-expert crew and low-expert crew, analyze their emotions, and incorporate them into the training plan."

[1406] This series of processes makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, develop specific training plans that take user emotions into consideration, and provide effective training support.

[1407] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1408] Step 1:

[1409] Data entry

[1410] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[1411] Input: Talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf)

[1412] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1413] Output: Text data sent to the server

[1414] Specific actions:

[1415] The user selects a talk paper from their PC or smartphone and clicks the upload button. The device reads the file and sends it to the server as text data.

[1416] Step 2:

[1417] Data storage and preprocessing

[1418] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[1419] Input: Text data sent to the server

[1420] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[1421] Output: Preprocessed text data

[1422] Specific actions:

[1423] The server saves the talk paper to the database, recording metadata such as the file name, upload date and time, and user ID. Next, the server uses a text analysis program to perform tokenization, stop word removal, and special character removal to generate pre-processed text data.

[1424] Step 3:

[1425] Keyword and phrase extraction

[1426] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from the pre-processed talk paper.

[1427] Input: Preprocessed text data

[1428] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[1429] Output: List of keywords and phrases and their frequencies

[1430] Specific actions:

[1431] The server applies natural language processing technology to extract important keywords (e.g., "customer satisfaction," "prompt response") and phrases from the talk paper. It calculates the frequency of occurrence of the extracted keywords and stores the results in a database.

[1432] Step 4:

[1433] Data comparison and analysis

[1434] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1435] Input: A list of keywords and phrases and their frequency of occurrence.

[1436] The server uses cosine similarity to calculate the similarity between the two talk papers and quantifies the discrepancy.

[1437] Output: Quantified deviation data

[1438] Specific actions:

[1439] The server applies the TF-IDF algorithm to calculate the importance of keywords within the talk paper. Next, it calculates cosine similarity to quantify the discrepancy between the talk papers of the high-expert crew and the low-crew crew.

[1440] Step 5:

[1441] Application of the emotion engine

[1442] The server uses an emotion engine (e.g., Google Cloud Natural Language API or IBM Watson) to analyze the user's emotions from keywords and phrases within the talk paper.

[1443] Input: Preprocessed text data

[1444] The server classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[1445] Output: Sentiment analysis result data

[1446] Specific actions:

[1447] The server uses an emotion engine to analyze keywords and phrases within the talk paper and determine an emotion score and emotion type. The results are then stored in a database.

[1448] Step 6:

[1449] Visualization of results

[1450] The server visualizes quantified deviation data and sentiment data.

[1451] Input: Quantified deviation data and sentiment analysis result data

[1452] The server uses a graphing library (e.g., D3.js or Chart.js) to visualize this data in bar graphs or radar charts.

[1453] Output: Visualized graphs and charts

[1454] Specific actions:

[1455] The server generates graphs based on deviation data and sentiment data, prepares data for display in a visually easy-to-understand format, and sends this data to the terminal.

[1456] Step 7:

[1457] Development and management of training plans

[1458] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[1459] Input: Visualized graphs and charts

[1460] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[1461] Output: Training plan and progress management data

[1462] Specific actions:

[1463] The user views the displayed graph data, creates a specific training plan for the low-level crew member, and saves it to the server. The server then takes sentiment data into consideration and automatically adjusts the training plan, setting up relevant reminders and notifications.

[1464] (Application Example 2)

[1465] Next, we will explain application example 2. In the following explanation, 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."

[1466] Previous methods for evaluating staff performance and developing training plans relied heavily on subjective assessments, making effective training difficult. Furthermore, failing to consider staff emotional states during training planning could lead to decreased motivation and stress. This invention aims to provide more effective training plans by quantifying the discrepancies between high-expert and low-level staff communication reports and by considering staff emotional states.

[1467] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1468] In this invention, the server includes means for inputting talk papers from high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to visualize quantified discrepancy data and sentiment data, formulate and manage training plans based on the results, and periodically monitor and notify progress. Furthermore, by including means for analyzing the user's emotions from keywords and phrases in the talk papers using an emotion analysis engine, it becomes possible to formulate training plans that reflect the emotional state of the staff.

[1469] A "high-expert crew" is a staff member who possesses high levels of experience and skills and demonstrates outstanding performance in their work.

[1470] A "low-level crew" refers to staff members who have relatively low experience and skills, and who still have room for improvement in their work.

[1471] A "talk paper" refers to a document or script that staff members use when responding to customers or giving explanations.

[1472] "Means of input" refers to devices or interfaces used to import staff members' talk papers into the system.

[1473] "Means of storage" refers to storage or databases used to record and retain entered data.

[1474] "Methods for preprocessing" refer to processes such as text tokenization, stop word removal, and special character deletion performed to prepare the input data into a format that is easy to analyze.

[1475] "Means for extracting keywords and phrases" refers to methods for extracting important words and phrases using natural language processing algorithms.

[1476] "Methods for quantifying discrepancies" refer to algorithms and analytical methods for quantitatively evaluating and expressing the differences between talk papers from high-expert crews and low-expert crews as numerical values.

[1477] "Means of visualization" refer to tools and software used to display analysis results in graphical formats such as bar graphs and radar charts.

[1478] "Means for formulating and managing training plans" refers to systems and mechanisms for formulating staff training plans, monitoring their progress, and issuing necessary notifications and reminders.

[1479] An "emotion analysis engine" refers to algorithms and technologies that read emotions from text data and classify them as positive, negative, neutral, etc.

[1480] "Monitoring and notification mechanisms" refer to functions that allow the system to periodically check the progress of the training plan and generate reminders and alerts as needed.

[1481] This invention consists of a system for inputting talk papers from high-expert crew members and low-level crew members, quantifying the discrepancies between them, and formulating and managing staff training plans. In addition, by using an emotion analysis engine to analyze the user's emotions and reflecting them in the training plan, it is possible to support more effective training.

[1482] Hardware and software to be used

[1483] Hardware:

[1484] Server: Used for processing and storing data.

[1485] Device: An input device such as a smartphone or smart glasses.

[1486] software:

[1487] Speech recognition software (Google Speech Recognition): Converts spoken conversation into text data.

[1488] Natural language processing libraries (nltk, scikit-learn): Used for text tokenization, stop word removal, special character removal, and keyword extraction.

[1489] Database: Used to store talk papers and analysis results.

[1490] Emotion analysis engine: Recognizes and analyzes emotions such as positive, negative, and neutral.

[1491] Program Processing Overview

[1492] Data entry and saving

[1493] Users upload talk papers for high-expert and low-level crews to the system using their smartphones or smart glasses. The server receives this data and stores it in a database. At the same time, it also records metadata (crew type, date and time, etc.).

[1494] Text preprocessing

[1495] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and special character deletion. This prepares the text for easier analysis.

[1496] Keyword and phrase extraction and comparison

[1497] The server uses natural language processing techniques to extract important keywords and phrases from pre-processed talk papers. Then, it evaluates the importance of the keywords using the TF-IDF algorithm, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[1498] Emotion analysis

[1499] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases in the talk paper. It classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[1500] Visualization of results

[1501] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts. This visualized data is sent to the terminal, and the results are displayed to the user.

[1502] Development and management of training plans

[1503] Based on the user's visualized results, the server develops a training plan for the low-crew member. The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it incorporates elements of mental support. The server supports the progress management of the training plan and provides necessary reminders and notifications.

[1504] Specific examples of implementation

[1505] For example, imagine a system where smart glasses are used to record conversations between staff working in a physical store and customers, and the data is analyzed. The recorded conversations are sent to a server and compared to the talk papers of high-expert crew members. Based on the analysis results, a staff training plan is developed. The progress of the plan is then monitored regularly, and notifications are sent to the user as needed. If the emotional data is negative, mental training is suggested.

[1506] Example of a prompt

[1507] "Please analyze the conversation text, calculate its similarity to advanced customer service skills, and perform sentiment analysis. The following conversation text: {Conversation Text}"

[1508] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[1509] Step 1:

[1510] Data entry

[1511] Users record the talk papers of high-expert and low-crew members using their smartphones or smart glasses and upload them to the system. The device converts the recorded audio data into text data and sends it to the server.

[1512] Input: Audio data

[1513] Output: Text data

[1514] Specific operation: Use speech recognition software (Google Speech Recognition) to convert speech data into text.

[1515] Step 2:

[1516] Data storage

[1517] The server stores the received text data in a database. At the same time, it also records metadata related to the talk paper (such as crew type and date / time).

[1518] Input: Text data, metadata

[1519] Output: Saved database entries

[1520] Specific operation: Connect to the database and save text data and metadata in a specified format.

[1521] Step 3:

[1522] Pre-treatment

[1523] The server retrieves talk papers from the database and performs preprocessing. Specifically, it performs tokenization, stop word removal, and removal of special characters.

[1524] Input: Text data

[1525] Output: Preprocessed text data

[1526] Specific operation: Uses the natural language processing library (nltk) to perform tokenization, stop word removal, and special character removal.

[1527] Step 4:

[1528] Keyword and phrase extraction

[1529] The server extracts important keywords and phrases from the pre-processed text data.

[1530] Input: Preprocessed text data

[1531] Output: List of keywords and phrases

[1532] Specific operation: Use TfidfVectorizer to evaluate and extract keywords and phrases from text based on their importance.

[1533] Step 5:

[1534] Quantifying the discrepancy

[1535] The server uses the TF-IDF algorithm and cosine similarity to calculate the similarity between the talk papers of the high-expert crew and the low-crew, and quantifies the discrepancy.

[1536] Input: List of keywords and phrases

[1537] Output: Quantified deviation data

[1538] Specific operation: Calculates similarity using TfidfVectorizer and cosine_similarity.

[1539] Step 6:

[1540] Emotion analysis

[1541] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases within the talk paper.

[1542] Input: List of keywords and phrases

[1543] Output: Sentiment analysis results (positive, negative, neutral, etc.)

[1544] Specific operation: Classifies the sentiment of keywords and phrases using a proprietary sentiment analysis algorithm.

[1545] Step 7:

[1546] Visualization of results

[1547] The server visualizes quantified deviation data and sentiment data and sends the results to the terminal.

[1548] Input: Quantified deviation data, sentiment data

[1549] Output: Visualized data (graphs, charts, etc.)

[1550] Specific operation: Use matplotlib to generate and visualize bar graphs and radar charts.

[1551] Step 8:

[1552] Development and management of training plans

[1553] Users create training plans for their low-level crew members based on visualized results. The server automatically adjusts the training plan based on sentiment data, monitors progress, and provides notifications.

[1554] Input: Visualized data, training plan

[1555] Output: Adjusted training plan, progress notifications

[1556] Specific operation: The server adjusts the care plan based on sentiment data stored in the database, monitors progress, and generates reminders and notifications as needed.

[1557] The specific processing unit 290 transmits the result of the specific processing to the headset terminal 314. In the headset terminal 314, the control unit 46A causes the speaker 240 and display 343 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[1558] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[1559] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and specific processing may also be performed by the headset terminal 314.

[1560] [Fourth Embodiment]

[1561] Figure 7 shows an example of the configuration of the data processing system 410 according to the fourth embodiment.

[1562] As shown in Figure 7, the 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.

[1563] The data processing device 12 comprises a computer 22, a database 24, and a communication interface 26. The computer 22 is an example of a "computer" related to the technology of this disclosure. The computer 22 comprises a processor 28, RAM 30, and storage 32. The processor 28, RAM 30, and storage 32 are connected to a bus 34. The database 24 and the communication interface 26 are also connected to the bus 34. The communication interface 26 is connected to a network 54. An example of the network 54 is a WAN (Wide Area Network) and / or a LAN (Local Area Network).

[1564] The robot 414 includes a computer 36, a microphone 238, a speaker 240, a camera 42, a communication interface 44, and a controlled object 443. The computer 36 includes a processor 46, RAM 48, and storage 50. The processor 46, RAM 48, and storage 50 are connected to a bus 52. The microphone 238, speaker 240, camera 42, and controlled object 443 are also connected to the bus 52.

[1565] The microphone 238 receives voice signals from the user 20 and receives instructions from the user 20. The microphone 238 captures the voice signals from the user 20, converts the captured voice into audio data, and outputs it to the processor 46. The speaker 240 outputs audio according to the instructions from the processor 46.

[1566] Camera 42 is a small digital camera equipped with an optical system including a lens, aperture, and shutter, and an image sensor such as a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor or a CCD (Charge Coupled Device) image sensor, and captures images of the area around the user 20 (for example, an imaging range defined by a field of view equivalent to the width of a typical healthy person's field of vision).

[1567] Communication interface 44 is connected to network 54. Communication interfaces 44 and 26 are responsible for the exchange of various information between processor 46 and processor 28 via network 54. The exchange of various information between processor 46 and processor 28 using communication interfaces 44 and 26 is performed in a secure manner.

[1568] The controlled object 443 includes a display device, LEDs in the eyes, and motors that drive 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 robot 414's emotions can be expressed by controlling these motors. Furthermore, the robot 414's facial expressions can also be expressed by controlling the illumination state of the LEDs in its eyes.

[1569] Figure 8 shows an example of the main functions of the data processing device 12 and the robot 414. As shown in Figure 8, the data processing device 12 performs specific processing using the processor 28. The storage 32 stores the specific processing program 56.

[1570] The specific processing program 56 is an example of a "program" relating to the technology of this 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.

[1571] The storage 32 stores the data generation model 58 and the emotion identification model 59. The data generation model 58 and the emotion identification model 59 are used by the identification processing unit 290.

[1572] In robot 414, the processor 46 performs the reception output processing. The storage 50 stores the reception output program 60. 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 processing is realized by the processor 46 operating as a control unit 46A according to the reception output program 60 executed on the RAM 48.

[1573] Next, the specific processing performed by the specific processing unit 290 of the data processing device 12 will be described. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1574] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[1575] Program Processing Overview

[1576] 1. Data entry

[1577] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[1578] The device retrieves the uploaded talk paper and sends it to the server.

[1579] 2. Data Storage

[1580] The server saves the received talk papers to the database.

[1581] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[1582] 3. Text preprocessing

[1583] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization and stop word removal.

[1584] This formats the talk paper in a way that makes it easier to analyze.

[1585] 4. Extraction of keywords and phrases

[1586] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from the pre-processed talk paper.

[1587] The frequency of keyword occurrences is calculated, and this data is stored in a database.

[1588] 5. Data Comparison and Analysis

[1589] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews.

[1590] As a concrete example, we use TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[1591] Based on these analysis results, the gap between high-expert crews and low-expert crews will be quantified and evaluated.

[1592] 6. Visualization of Results

[1593] The server visualizes the quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a format that is easy for users to understand.

[1594] This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[1595] 7. Planning and management of training programs

[1596] Based on the results displayed by the system, users develop training plans for low-level crew members.

[1597] The server stores this training plan in a database and provides reminder and notification functions for progress management.

[1598] This allows for continuous tracking of progress towards achieving the training plan and enables adjustments to the plan as needed.

[1599] Specific example

[1600] Example 1:

[1601] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[1602] The server saves the talk paper to the database and performs preprocessing.

[1603] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[1604] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[1605] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[1606] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[1607] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[1608] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[1609] The following describes the processing flow.

[1610] Step 1:

[1611] The user accesses the system, enters their account information, and logs in.

[1612] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[1613] Step 2:

[1614] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[1615] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1616] Step 3:

[1617] The server saves the received talk papers to the database.

[1618] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[1619] Step 4:

[1620] The server passes the saved talk paper to the text analysis program.

[1621] The text analysis program performs preprocessing such as tokenization, stop word removal, and special character deletion.

[1622] Step 5:

[1623] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[1624] The frequency of occurrence of extracted keywords and phrases is calculated and stored in a database.

[1625] Step 6:

[1626] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1627] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[1628] Step 7:

[1629] The server visualizes the data in visual formats such as bar graphs and radar charts, based on the quantified deviation data.

[1630] Send the visualized data to the device.

[1631] Step 8:

[1632] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1633] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[1634] Step 9:

[1635] The user enters the training plan they have created into the system and sends it to the server.

[1636] The server saves the training plan to a database and sets up reminder and notification functions for progress management.

[1637] Step 10:

[1638] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[1639] (Example 1)

[1640] Next, we will describe Example 1. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1641] Conventional talk paper analysis systems have made it difficult to accurately grasp the performance differences between high-expert and low-level crews and to develop concrete training plans. Furthermore, there was a lack of systems to provide analysis results in a visually easy-to-understand format and to manage the progress of training plans based on those results. Therefore, the inability to achieve efficient and effective crew training remains a challenge.

[1642] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 1 is realized by the following means.

[1643] In this invention, the server includes means for a user to log in and input data on high-expert crew and low-crew; means for a terminal to acquire the input data and send it to the server; means for the server to store the received data in a database and record metadata; means for the server to perform preprocessing on the stored data, such as tokenization and stop word removal; means for the server to extract keywords and important phrases from the preprocessed data; means for the server to compare the frequency of occurrence and usage of the extracted keywords and phrases and quantify the discrepancies; means for the server to visualize the quantified discrepancy data, create graphs and charts and send them to the terminal; means for the user to formulate a training plan based on the visualized results and save it to the server; and means for the server to manage the progress of the training plan and provide reminder and notification functions. This makes it possible to compare high-expert crew and low-crew, and to formulate and manage specific training plans based on the results.

[1644] A "user" refers to a person who accesses the system, enters data, or checks the results.

[1645] "Terminal" refers to a device used by a user, such as a computer or smartphone.

[1646] A "server" refers to a central processing unit that stores, processes, and analyzes data.

[1647] A "database" refers to a system for systematically storing and managing data.

[1648] "Metadata" refers to additional information accompanying the main data, such as the data type and creation date and time.

[1649] "Preprocessing" refers to a series of preparatory tasks performed before data analysis, such as tokenization and stop word removal.

[1650] "Tokenization" refers to the process of dividing text into smaller units such as words and phrases.

[1651] A "stop word" refers to a word that is frequently used in text analysis but does not have any particular meaning.

[1652] A "keyword" refers to a word or short phrase that has significant meaning within a text.

[1653] A "phrase" is a part of language, referring to a combination of multiple words that have a specific meaning on their own.

[1654] "Frequency" refers to an indicator of how often a particular word or phrase is used within a text.

[1655] "Discrepancy" refers to the differences in performance and terminology used between high-expert crews and low-level crews.

[1656] "Quantification" refers to representing data or results using specific numerical values.

[1657] "Visualization" refers to the visual representation of data using graphs and charts.

[1658] A "development plan" refers to a plan designed to improve the skills and performance of low-level crew members.

[1659] A "reminder" refers to an alert that is sent at a specific date or time.

[1660] "Notification function" refers to a system function that transmits information to the user.

[1661] Modes for carrying out the invention

[1662] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system analyzes the talk papers entered by the user and provides visualized results, thereby supporting the efficient planning and management of training programs.

[1663] System Configuration

[1664] This system mainly consists of the following elements:

[1665] 1. User terminal

[1666] This is a device used by users to log in and upload talk papers. This includes personal computers, tablets, smartphones, etc.

[1667] 2. Server

[1668] It is a central processing unit that handles data storage, preprocessing, analysis, visualization, and management of development plans.

[1669] 3. Database

[1670] This is a storage device for saving talk papers, analysis results, and training plans. MongoDB will be used here.

[1671] Software to use

[1672] The software configuration of this system is mainly as follows:

[1673] 1. Text Analysis Library

[1674] NLTK (Natural Language Toolkit): Used for tokenizing text and removing stop words.

[1675] 2. Natural Language Processing (NLP) Libraries

[1676] SpaCy: Used to extract keywords and phrases from talk papers.

[1677] 3. Data Visualization Library

[1678] Matplotlib and Plotly are used to visualize quantified deviation data.

[1679] System operation

[1680] 1. Data entry

[1681] The user logs in and uploads the talk papers for the high-expert crew and the low-crew.

[1682] The terminal retrieves the talk paper and sends it to the server.

[1683] 2. Data Storage

[1684] The server saves the received talk papers to a database and records metadata (such as crew type and date / time).

[1685] 3. Text preprocessing

[1686] The server tokenizes the talk paper using NLTK and removes stop words.

[1687] 4. Extraction of keywords and phrases

[1688] The server uses SpaCy to extract keywords and important phrases from the pre-processed talk paper. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[1689] 5. Data Comparison and Analysis

[1690] The server compares the frequency of keywords and phrases used by high-expert crews and low-expert crews.

[1691] The server uses TF-IDF to calculate the importance of each keyword and cosine similarity to measure the similarity between talk papers.

[1692] 6. Visualization of Results

[1693] The server visualizes the quantified deviation data using Matplotlib or Plotly to create bar graphs and radar charts.

[1694] The device displays the visualized results to the user.

[1695] 7. Planning and management of training programs

[1696] Based on the results visualized by the user, a training plan for the low-level crew is developed and saved on the server.

[1697] The server manages the progress of the training plan and provides reminders and notifications.

[1698] Specific example

[1699] Example 1:

[1700] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[1701] The server saves the talk paper to the database and performs preprocessing.

[1702] The server uses NLP (Neuro-Linguistic Programming) techniques to extract keywords from the talk papers of Crew A and Crew B.

[1703] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[1704] The server uses the digitized data to create bar graphs and radar charts, which are then sent to the terminal for display to the user.

[1705] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[1706] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[1707] This makes it possible to objectively evaluate the performance difference between high-expert and low-level crews and to develop specific training plans.

[1708] Example of a prompt

[1709] "Analyze the following talk paper and quantify the performance gap between the high-expert crew and the low-expert crew."

[1710] The flow of the specific processing in Example 1 will be explained using Figure 11.

[1711] Step 1:

[1712] The user logs into the system and uploads their talk papers for both the high-expert crew and the low-crew. The inputs here are the user's login information and the talk paper files. The terminal receives this information and sends it to the server. The login information is verified, and if successful, the talk paper upload screen appears. The user clicks the upload button, selects the file, and submits it.

[1713] Step 2:

[1714] The server saves the received talk paper to a database and records metadata (e.g., crew type, upload date and time). The input for this step is the talk paper file and metadata sent from the terminal. The server temporarily stores the file and converts it to a format suitable for storage in the database. It then saves it to the database along with the metadata. The output is the talk paper and metadata stored in the database.

[1715] Step 3:

[1716] The server preprocesses the stored talk papers using the NLTK library. The input for this step is the talk papers loaded from the database. Specifically, the server tokenizes the text (divides it into words) and removes stop words (e.g., "and," "the," etc.). This process formats the talk papers in a way that is suitable for analysis. The output is the preprocessed text data.

[1717] Step 4:

[1718] The server extracts keywords and important phrases from pre-processed text data using the SpaCy library. The input for this step is the pre-processed text data. The server identifies important phrases such as noun phrases and verb phrases and calculates their frequency. Specifically, the server calculates the frequency of occurrence of each keyword and stores it in the database. The output is data of the extracted keywords and their frequencies.

[1719] Step 5:

[1720] The server compares the frequency of keywords and phrases between high-expert and low-expert crews. The input for this step is keyword and phrase frequency data stored in a database. The server calculates TF-IDF to assess the importance of each keyword. It also uses cosine similarity to measure the similarity between talk papers. Specifically, the server quantifies the comparison results and stores them in a database. The output is quantified deviation data.

[1721] Step 6:

[1722] The server visualizes the quantified deviation data using Matplotlib or Plotly. The input for this step is the quantified deviation data. The server creates graphs in visually easy-to-understand formats such as bar graphs and radar charts. Specifically, the server generates graph data and sends it to the terminal. The output is the visualized graph data.

[1723] Step 7:

[1724] The terminal displays the visualized results to the user. The input for this step is the visualized graph data sent from the server. The terminal displays the graph on the screen so that the user can view it. Specifically, the graph is displayed in the terminal's web browser or similar application. The output is the graph displayed to the user.

[1725] Step 8:

[1726] The user develops a training plan for low-level crew members based on the visualized results. The input for this step is visualized graph data. The user fills in a specific training plan in an input form and sends it to the server. Specifically, the user decides on a training plan and enters it into the input form. The output is the training plan data sent to the server.

[1727] Step 9:

[1728] The server stores the training plan in a database and provides reminder and notification functions for progress management. The input for this step is the training plan data submitted by the user. The server stores the training plan in the database and sets reminders and notifications based on the schedule. Specifically, the server sets the notification schedule and sends reminders to the user. The output is the reminders and notifications sent to the user.

[1729] (Application Example 1)

[1730] Next, we will explain Application Example 1. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1731] Traditionally, the efficiency improvements and training plans for robot operations within factories were primarily relied upon human experience and intuition. This resulted in insufficient transfer of knowledge and skills from experienced operators to new recruits, leading to decreased productivity and increased operational errors. Furthermore, the lack of concrete methods for incorporating the results of talk paper analysis into robot implementation made overall optimization difficult.

[1732] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 1 is realized by the following means.

[1733] In this invention, the server includes means for inputting talk papers of high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to quantify the discrepancies between talk papers and efficiently plan and manage training programs. Furthermore, by linking it with a robot that optimizes operations, it is possible to immediately reflect the analysis results and improve the efficiency of the entire factory.

[1734] A "high-expert crew" is a group of operators who possess advanced expertise and skills in specific tasks or operations.

[1735] A "low crew" is a group of operators who are still in training or have low skill levels in a particular task or job.

[1736] A "talk paper" is a set of procedures, manuals, or customer service guidelines that operators refer to during their work.

[1737] "Input method" refers to the specific hardware or software interface used to input data into a system.

[1738] "Means of preservation" refer to storage devices and database systems for retaining data over the long term.

[1739] "Means of preprocessing" refers to the process of shaping and processing data to facilitate data analysis, and the program that executes it.

[1740] "Means for extracting keywords and phrases" refer to algorithms and techniques for identifying important words and phrases from text data.

[1741] "Methods for quantifying discrepancies" refer to methods and algorithms for measuring the differences between different datasets and expressing them numerically.

[1742] "Means of visualization and displaying results" refers to software or tools for displaying analyzed data in a visual form, such as graphs or charts.

[1743] "Means for formulating and managing training plans" refers to systems and methodologies for developing, tracking, and managing plans for improving operators' skills.

[1744] "Means for linking with a robot and displaying analysis results" refers to an interface and technology for integrating analyzed data into the control system of an industrial robot and displaying the results in real time.

[1745] This invention relates to a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies between them. This system is intended to optimize the operation of robots operating in a factory and is implemented in the following steps.

[1746] System Configuration

[1747] Data entry and saving:

[1748] The user first uploads the talk papers for the high-expert and low-expert crews to the system. The terminal retrieves these talk papers and sends them to the server. The server saves the received talk papers to its database. During this saving process, metadata such as the crew type and date and time are also recorded.

[1749] Text preprocessing and parsing:

[1750] The server first performs preprocessing on the talk papers stored in the database before analyzing them. This preprocessing includes tokenization and stop word removal. Through this process, the talk papers are formatted to be easily analyzed. Furthermore, natural language processing (NLP) techniques are used to extract keywords and important phrases from the preprocessed talk papers. Specific examples include using TF-IDF (Term Frequency-Inverse Document Frequency) to evaluate keyword importance and using cosine similarity to measure the similarity between talk papers.

[1751] Data comparison and analysis:

[1752] The server compares the frequency of keyword occurrences and phrase usage between high-expert crews and low-expert crews. Based on this analysis, the discrepancy between high-expert and low-expert crews is quantified and evaluated. The quantified data is stored in a database.

[1753] Visualization and display:

[1754] The server visualizes quantified deviation data. For example, it creates bar graphs and radar charts and displays them in a user-friendly format. This visualization allows users to quickly grasp specific areas for improvement and areas that need development.

[1755] Development and management of training plans:

[1756] Based on the results displayed by the system, users create training plans for their low-level crew members. The server stores these training plans in a database and provides reminders and notifications for progress management. This allows for continuous tracking of the training plan's progress and adjustments to the plan as needed.

[1757] Robot collaboration and optimization:

[1758] The server transmits the analysis results to the industrial robots. This allows for real-time display of the analysis results and optimization of operations. This integration supports the overall efficiency of the factory.

[1759] Specific example

[1760] For example, if a high-expert crew uses a talk paper that emphasizes "building trust for customer satisfaction and quick response," and a low-crew uses a talk paper that emphasizes "building trust with customers for quick response," the server compares these talk papers. First, it extracts keywords from each talk paper and calculates the TF-IDF value. Then, it calculates cosine similarity to quantify the divergence between the two talk papers. The quantified data is visualized in bar graphs or radar charts and presented to the user.

[1761] Examples of prompts for generative AI models

[1762] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[1763] The flow of a specific process in Application Example 1 will be explained using Figure 12.

[1764] Step 1:

[1765] The user logs into the system and uploads talk papers for both the high-expert crew and the low-crew. The terminal retrieves these talk papers and sends them to the server. The talk papers are expected to be in file formats such as text files or PDFs.

[1766] Input: Talk papers for the high-expert crew and the low-crew crew (text or PDF format).

[1767] Output: Talk paper sent to the server.

[1768] Step 2:

[1769] The server saves uploaded talk papers to the database. Along with the talk papers, metadata (such as crew type and upload date / time) is also saved. This metadata will be useful for future data analysis.

[1770] Input: Talk paper and metadata sent to the server.

[1771] Output: Talk papers and metadata stored in the database.

[1772] Step 3:

[1773] The server retrieves the talk papers stored in the database for preprocessing. Preprocessing includes tokenization, stop word removal, and normalization. These preprocessing steps prepare the text data for easier analysis.

[1774] Input: Talk paper retrieved from the database.

[1775] Output: Preprocessed text data.

[1776] Step 4:

[1777] The server uses natural language processing (NLP) techniques to extract keywords and important phrases from pre-processed talk papers. Specifically, TF-IDF (Term Frequency-Inverse Document Frequency) is used.

[1778] Input: Preprocessed text data.

[1779] Output: Extracted keywords and phrases, and their frequency data.

[1780] Step 5:

[1781] The server uses TF-IDF to evaluate the importance of each keyword and then uses cosine similarity to measure the similarity between talk papers from high-expert and low-expert crews. This analysis quantifies the discrepancies between talk papers.

[1782] Input: Keywords and phrases, and their frequency data.

[1783] Output: Numerical deviation data.

[1784] Step 6:

[1785] The server visualizes quantified deviation data. It creates bar graphs and radar charts, displaying the results in a user-friendly format. This allows users to visually identify areas that require improvement.

[1786] Input: Numerical deviation data.

[1787] Output: Results visualized as bar graphs or radar charts.

[1788] Step 7:

[1789] The user develops a training plan for the low-level crew based on the results displayed within the system. The server stores this training plan in a database and provides reminders and notifications for progress management.

[1790] Input: Visualized result.

[1791] Output: Training plan stored in the database.

[1792] Step 8:

[1793] The server links the analysis results to the control system of the industrial robot. This linkage allows the analysis results to be reflected in the robot's operation in real time, optimizing the work process.

[1794] Input: Analysis results.

[1795] Output: Analysis results reflected in the robot.

[1796] Specific examples of operation

[1797] If a senior operator's talk paper states "We prioritize trust for customer satisfaction and prompt response," and a junior operator's talk paper states "We build trust with customers to respond quickly," the server analyzes these talk papers. It extracts keywords, compares their frequencies, and derives several characteristics. Then, it quantifies the discrepancies, visualizes them, and provides them to the user. Based on these results, the user creates a training plan, which the server saves to a database. Finally, the analysis results are linked to a robot to optimize operations.

[1798] Examples of prompts for generative AI models

[1799] We will compare the talk papers of high-expert crew members with those of low-level crew members and quantify the discrepancies in keywords and phrases. Furthermore, we will visualize the results and build a system for planning and managing the training of low-level crew members.

[1800] Furthermore, an emotion engine that estimates the user's emotions may be incorporated. That is, the identification processing unit 290 may use the emotion identification model 59 to estimate the user's emotions and perform identification processing using the user's emotions.

[1801] This invention combines a system that compares the talk papers of high-expert crews and low-expert crews and quantifies the discrepancies with an emotion engine. This system not only analyzes the talk papers entered by users and quantifies the discrepancies, but also recognizes and analyzes the user's emotions and reflects them in the training plan, thereby supporting more effective training.

[1802] Program Processing Overview

[1803] 1. Data entry

[1804] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[1805] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1806] 2. Data Storage

[1807] The server saves the received talk papers to the database.

[1808] When saving a talk paper, metadata (such as crew type and date / time) is also recorded.

[1809] 3. Text preprocessing

[1810] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[1811] 4. Extraction of keywords and phrases

[1812] The server uses natural language processing (NLP) techniques to extract important keywords and phrases from the pre-processed talk paper.

[1813] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[1814] 5. Data Comparison and Analysis

[1815] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1816] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[1817] 6. Application of the Emotion Engine

[1818] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[1819] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[1820] 7. Visualization of Results

[1821] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[1822] Send the visualized data to the device.

[1823] 8. Planning and management of training programs

[1824] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[1825] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[1826] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[1827] Specific example

[1828] Example 1:

[1829] The user logs into the system and uploads the talk papers for High Expert Crew A and Low Crew B.

[1830] The server saves the talk paper to the database and performs preprocessing.

[1831] The server uses NLP technology to extract keywords (for example, "customer satisfaction," "prompt response," and "trust") from the talk papers of crew A and crew B.

[1832] The server uses TF-IDF and cosine similarity to quantify and evaluate the discrepancies between the talk papers of crew A and crew B.

[1833] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" frequently appear in Crew B's talk paper, the user's emotion will be classified as "negative."

[1834] The server visualizes quantified data and sentiment data in bar graphs and radar charts, and also displays sentiment-related indicators.

[1835] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1836] Based on the results displayed to the user, a training plan for Low Crew B is devised and saved to the server.

[1837] The server automatically adjusts the training plan based on emotional data, adding elements such as mental training and counseling.

[1838] The server supports the management of the training plan's progress and sends necessary reminders and notifications to the user.

[1839] This makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, and to create concrete training plans that take into account user sentiment.

[1840] The following describes the processing flow.

[1841] Step 1:

[1842] The user accesses the system, enters their account information, and logs in.

[1843] The server verifies the user's authentication information, and if authentication is successful, it displays the dashboard screen.

[1844] Step 2:

[1845] Users select and upload talk papers for both the high-expert crew and the low-crew crew.

[1846] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1847] Step 3:

[1848] The server saves the received talk papers to the database.

[1849] When saving, metadata such as the type of crew the talk paper is created for and the upload date and time are also recorded.

[1850] Step 4:

[1851] The server passes the saved talk paper to the text analysis program.

[1852] The text analysis program performs preprocessing such as tokenization, stop word removal, and removal of special characters.

[1853] Step 5:

[1854] The server extracts important keywords and phrases from pre-processed text data using natural language processing (NLP) techniques.

[1855] The frequency of occurrence of extracted keywords and phrases is calculated, and this data is stored in a database.

[1856] Step 6:

[1857] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1858] The similarity between the two talk papers is calculated using cosine similarity, and the discrepancy is quantified.

[1859] Step 7:

[1860] The server uses an emotion engine to analyze the user's emotions from keywords and phrases within the talk paper.

[1861] The system classifies users' emotions into positive, negative, neutral, etc., and stores the results in a database.

[1862] Step 8:

[1863] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts.

[1864] Send the visualized data to the device.

[1865] Step 9:

[1866] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1867] Based on the results displayed to the user, a training plan for low-level crew members will be developed.

[1868] Step 10:

[1869] The user enters the training plan they have created into the system and sends it to the server.

[1870] The server stores the training plan in a database and provides reminder and notification functions for progress management.

[1871] Step 11:

[1872] The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it will incorporate elements of mental support.

[1873] Step 12:

[1874] The server periodically sends reminders and notifications to users to support the progress of their training plan.

[1875] (Example 2)

[1876] Next, we will describe Example 2. In the following description, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1877] Conventional training planning systems made it difficult to compare the performance of high-expert crew members with that of low-level crew members, and they were unable to create specific training plans that took into account user emotions. Furthermore, there was a lack of means to quantify discrepancies in text documents and analyze and visualize emotional data, which hindered effective training support.

[1878] The identification process performed by the identification processing unit 290 of the data processing device 12 in Example 2 is realized by the following means.

[1879] In this invention, the server includes means for inputting text documents of high-expert crew and low-crew; means for saving and pre-processing the input text documents; means for extracting keywords and phrases from the pre-processed text documents; means for comparing the frequency and context of the extracted keywords and phrases and quantifying the discrepancies; means for analyzing sentiment from the keywords and phrases and saving the results; means for visualizing the quantified discrepancy data and sentiment data and displaying the results; and means for formulating and managing training plans based on the displayed results. This makes it possible to objectively evaluate the performance differences between high-expert crew and low-crew, formulate specific training plans that take into account user sentiment, and provide effective training support.

[1880] A "text document" refers to a document file saved in text format, including, for example, PDF, Word, and Excel files.

[1881] "Input method" refers to the interface that allows users to upload text documents to the system, and it uses specific software and hardware.

[1882] "Means of saving and pre-processing" refers to the process by which the server saves the received text document to a database and then performs text processing such as tokenization, stop word removal, and removal of special characters.

[1883] "Means for extracting keywords and phrases" refers to the process of identifying and extracting important words and phrases from a text document using natural language processing algorithms.

[1884] "Methods for comparing frequency of occurrence and context and quantifying the discrepancies" refers to the process of analyzing the frequency of occurrence and contextual information of extracted keywords and phrases, and expressing the differences between high-expert crews and low-level crews numerically.

[1885] "Methods for analyzing emotions and saving results" refers to the process of using an emotion engine to identify a user's emotions from keywords and phrases within a text document and recording the results in a database.

[1886] "Means of visualization and displaying results" refers to an interface that visually represents quantified deviation data and sentiment data in the form of graphs and charts, and displays them to the user.

[1887] "Means for formulating and managing training plans" refers to the function of formulating training policies for low-level employees based on visualized data, and tracking and managing their progress.

[1888] This invention combines a system that analyzes text documents (hereinafter referred to as "talk papers") of high-expert crews and low-level crews and quantifies the discrepancies between them with an emotion engine. This system aims to support more effective training by not only analyzing the talk papers entered by users and quantifying the discrepancies, but also recognizing and analyzing the user's emotions and reflecting them in the training plan.

[1889] Hardware and software to be used

[1890] The system consists of the following main hardware and software components.

[1891] Device: A device such as a PC, smartphone, or tablet used by the user to upload their talk paper. Web applications or mobile applications are used.

[1892] Server: A central processing unit for data storage, preprocessing, analysis, and visualization. Server software used includes Apache and Nginx, among others.

[1893] Database: A relational database management system (RDBMS) used to store talk papers and analysis results. MySQL and PostgreSQL are commonly used.

[1894] NLP libraries: Software libraries for natural language processing. These include NLTK, SpaCy, BERT, Word2Vec, and others.

[1895] Emotion engine: APIs or software used for sentiment analysis. Examples include Google Cloud Natural Language API and IBM Watson.

[1896] Graph libraries: Tools for visualizing data. D3.js and Chart.js are commonly used.

[1897] Specific processing flow

[1898] 1. Data entry

[1899] Users log into the system and upload the talk papers for their high-expert and low-expert crews. For example, a user selects the talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf) from their PC or smartphone and clicks the upload button.

[1900] 2. Data storage and preprocessing

[1901] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[1902] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[1903] 3. Keyword and phrase extraction

[1904] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from pre-processed talk papers. For example, it might extract keywords such as "customer satisfaction," "prompt response," and "trust."

[1905] 4. Data Comparison and Analysis

[1906] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[1907] The server uses the TF-IDF algorithm to evaluate the importance of keywords, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[1908] 5. Application of the Emotion Engine

[1909] The server uses an emotion engine to analyze the user's emotions from keywords and phrases in the talk paper. For example, if negative keywords such as "difficulty" and "failure" appear frequently, the user's emotion will be classified as "negative."

[1910] 6. Visualization of Results

[1911] The server visualizes quantified deviation and sentiment data. In addition to bar graphs and radar charts, it displays sentiment-related indicators.

[1912] The device displays the visualized results to the user, clearly indicating specific areas for improvement and areas that need further development.

[1913] 7. Planning and management of training programs

[1914] Based on the results visualized by the user, a training plan for low-level crew members will be developed. For example, "mental training courses" and "communication skills improvement sessions" will be considered.

[1915] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[1916] A possible example of a specific prompt might be, "Extract keywords from the talk papers of the high-expert crew and low-expert crew, analyze their emotions, and incorporate them into the training plan."

[1917] This series of processes makes it possible to objectively evaluate the performance differences between high-expert and low-level crews, develop specific training plans that take user emotions into consideration, and provide effective training support.

[1918] The flow of the specific processing in Example 2 will be explained using Figure 13.

[1919] Step 1:

[1920] Data entry

[1921] The user logs into the system and uploads the talk papers for the high-expert crew and the low-crew.

[1922] Input: Talk paper files (e.g., crewA_talk.pdf, crewB_talk.pdf)

[1923] The device retrieves the uploaded talk paper as text data and sends it to the server.

[1924] Output: Text data sent to the server

[1925] Specific actions:

[1926] The user selects a talk paper from their PC or smartphone and clicks the upload button. The device reads the file and sends it to the server as text data.

[1927] Step 2:

[1928] Data storage and preprocessing

[1929] The server saves the received talk papers to the database. Metadata (e.g., crew type, date and time, user ID, etc.) is added during saving.

[1930] Input: Text data sent to the server

[1931] The server passes the saved talk papers to a text analysis program (e.g., NLTK or SpaCy) for preprocessing such as tokenization, stop word removal, and special character removal.

[1932] Output: Preprocessed text data

[1933] Specific actions:

[1934] The server saves the talk paper to the database, recording metadata such as the file name, upload date and time, and user ID. Next, the server uses a text analysis program to perform tokenization, stop word removal, and special character removal to generate pre-processed text data.

[1935] Step 3:

[1936] Keyword and phrase extraction

[1937] The server uses natural language processing techniques (e.g., BERT and Word2Vec) to extract important keywords and phrases from the pre-processed talk paper.

[1938] Input: Preprocessed text data

[1939] The server calculates the frequency of occurrence of extracted keywords and phrases and stores that data in a database.

[1940] Output: List of keywords and phrases and their frequencies

[1941] Specific actions:

[1942] The server applies natural language processing technology to extract important keywords (e.g., "customer satisfaction," "prompt response") and phrases from the talk paper. It calculates the frequency of occurrence of the extracted keywords and stores the results in a database.

[1943] Step 4:

[1944] Data comparison and analysis

[1945] The server uses the TF-IDF algorithm to evaluate the importance of keywords in the talk papers of the high-expert crew and the low-crew.

[1946] Input: A list of keywords and phrases and their frequency of occurrence.

[1947] The server uses cosine similarity to calculate the similarity between the two talk papers and quantifies the discrepancy.

[1948] Output: Quantified deviation data

[1949] Specific actions:

[1950] The server applies the TF-IDF algorithm to calculate the importance of keywords within the talk paper. Next, it calculates cosine similarity to quantify the discrepancy between the talk papers of the high-expert crew and the low-crew crew.

[1951] Step 5:

[1952] Application of the emotion engine

[1953] The server uses an emotion engine (e.g., Google Cloud Natural Language API or IBM Watson) to analyze the user's emotions from keywords and phrases within the talk paper.

[1954] Input: Preprocessed text data

[1955] The server classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[1956] Output: Sentiment analysis result data

[1957] Specific actions:

[1958] The server uses an emotion engine to analyze keywords and phrases within the talk paper and determine an emotion score and emotion type. The results are then stored in a database.

[1959] Step 6:

[1960] Visualization of results

[1961] The server visualizes quantified deviation data and sentiment data.

[1962] Input: Quantified deviation data and sentiment analysis result data

[1963] The server uses a graphing library (e.g., D3.js or Chart.js) to visualize this data in bar graphs or radar charts.

[1964] Output: Visualized graphs and charts

[1965] Specific actions:

[1966] The server generates graphs based on deviation data and sentiment data, prepares data for display in a visually easy-to-understand format, and sends this data to the terminal.

[1967] Step 7:

[1968] Development and management of training plans

[1969] Based on the results visualized by the user, a training plan for the low-level crew is developed.

[1970] Input: Visualized graphs and charts

[1971] The server automatically adjusts the training plan based on emotional data and saves the plan to a database. It also sets up email notifications as reminders and periodically checks on the progress.

[1972] Output: Training plan and progress management data

[1973] Specific actions:

[1974] The user views the displayed graph data, creates a specific training plan for the low-level crew member, and saves it to the server. The server then takes sentiment data into consideration and automatically adjusts the training plan, setting up relevant reminders and notifications.

[1975] (Application Example 2)

[1976] Next, we will explain application example 2. In the following explanation, the data processing device 12 will be referred to as the "server" and the robot 414 as the "terminal".

[1977] Previous methods for evaluating staff performance and developing training plans relied heavily on subjective assessments, making effective training difficult. Furthermore, failing to consider staff emotional states during training planning could lead to decreased motivation and stress. This invention aims to provide more effective training plans by quantifying the discrepancies between high-expert and low-level staff communication reports and by considering staff emotional states.

[1978] The specific processing performed by the specific processing unit 290 of the data processing device 12 in Application Example 2 is realized by the following means.

[1979] In this invention, the server includes means for inputting talk papers from high-expert crew and low-crew, means for saving and pre-processing the input talk papers, and means for extracting keywords and phrases from the pre-processed talk papers. This makes it possible to visualize quantified discrepancy data and sentiment data, formulate and manage training plans based on the results, and periodically monitor and notify progress. Furthermore, by including means for analyzing the user's emotions from keywords and phrases in the talk papers using an emotion analysis engine, it becomes possible to formulate training plans that reflect the emotional state of the staff.

[1980] A "high-expert crew" is a staff member who possesses high levels of experience and skills and demonstrates outstanding performance in their work.

[1981] A "low-level crew" refers to staff members who have relatively low experience and skills, and who still have room for improvement in their work.

[1982] A "talk paper" refers to a document or script that staff members use when responding to customers or giving explanations.

[1983] "Means of input" refers to devices or interfaces used to import staff members' talk papers into the system.

[1984] "Means of storage" refers to storage or databases used to record and retain entered data.

[1985] "Methods for preprocessing" refer to processes such as text tokenization, stop word removal, and special character deletion performed to prepare the input data into a format that is easy to analyze.

[1986] "Means for extracting keywords and phrases" refers to methods for extracting important words and phrases using natural language processing algorithms.

[1987] "Methods for quantifying discrepancies" refer to algorithms and analytical methods for quantitatively evaluating and expressing the differences between talk papers from high-expert crews and low-expert crews as numerical values.

[1988] "Means of visualization" refer to tools and software used to display analysis results in graphical formats such as bar graphs and radar charts.

[1989] "Means for formulating and managing training plans" refers to systems and mechanisms for formulating staff training plans, monitoring their progress, and issuing necessary notifications and reminders.

[1990] An "emotion analysis engine" refers to algorithms and technologies that read emotions from text data and classify them as positive, negative, neutral, etc.

[1991] "Monitoring and notification mechanisms" refer to functions that allow the system to periodically check the progress of the training plan and generate reminders and alerts as needed.

[1992] This invention consists of a system for inputting talk papers from high-expert crew members and low-level crew members, quantifying the discrepancies between them, and formulating and managing staff training plans. In addition, by using an emotion analysis engine to analyze the user's emotions and reflecting them in the training plan, it is possible to support more effective training.

[1993] Hardware and software to be used

[1994] Hardware:

[1995] Server: Used for processing and storing data.

[1996] Device: An input device such as a smartphone or smart glasses.

[1997] software:

[1998] Speech recognition software (Google Speech Recognition): Converts spoken conversation into text data.

[1999] Natural language processing libraries (nltk, scikit-learn): Used for text tokenization, stop word removal, special character removal, and keyword extraction.

[2000] Database: Used to store talk papers and analysis results.

[2001] Emotion analysis engine: Recognizes and analyzes emotions such as positive, negative, and neutral.

[2002] Program Processing Overview

[2003] Data entry and saving

[2004] Users upload talk papers for high-expert and low-level crews to the system using their smartphones or smart glasses. The server receives this data and stores it in a database. At the same time, it also records metadata (crew type, date and time, etc.).

[2005] Text preprocessing

[2006] The server passes the talk paper to a text analysis program, which performs preprocessing such as tokenization, stop word removal, and special character deletion. This prepares the text for easier analysis.

[2007] Keyword and phrase extraction and comparison

[2008] The server uses natural language processing techniques to extract important keywords and phrases from pre-processed talk papers. Then, it evaluates the importance of the keywords using the TF-IDF algorithm, calculates the similarity between the two talk papers using cosine similarity, and quantifies the discrepancy.

[2009] Emotion analysis

[2010] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases in the talk paper. It classifies the user's emotions as positive, negative, neutral, etc., and stores the results in a database.

[2011] Visualization of results

[2012] The server visualizes quantified deviation and sentiment data. For example, it displays sentiment-related indicators in addition to bar graphs and radar charts. This visualized data is sent to the terminal, and the results are displayed to the user.

[2013] Development and management of training plans

[2014] Based on the user's visualized results, the server develops a training plan for the low-crew member. The server automatically adjusts the training plan based on emotional data. For example, if there are many negative emotions, it incorporates elements of mental support. The server supports the progress management of the training plan and provides necessary reminders and notifications.

[2015] Specific examples of implementation

[2016] For example, imagine a system where smart glasses are used to record conversations between staff working in a physical store and customers, and the data is analyzed. The recorded conversations are sent to a server and compared to the talk papers of high-expert crew members. Based on the analysis results, a staff training plan is developed. The progress of the plan is then monitored regularly, and notifications are sent to the user as needed. If the emotional data is negative, mental training is suggested.

[2017] Example of a prompt

[2018] "Please analyze the conversation text, calculate its similarity to advanced customer service skills, and perform sentiment analysis. The following conversation text: {Conversation Text}"

[2019] The flow of a specific process in Application Example 2 will be explained using Figure 14.

[2020] Step 1:

[2021] Data entry

[2022] Users record the talk papers of high-expert and low-crew members using their smartphones or smart glasses and upload them to the system. The device converts the recorded audio data into text data and sends it to the server.

[2023] Input: Audio data

[2024] Output: Text data

[2025] Specific operation: Use speech recognition software (Google Speech Recognition) to convert speech data into text.

[2026] Step 2:

[2027] Data storage

[2028] The server stores the received text data in a database. At the same time, it also records metadata related to the talk paper (such as crew type and date / time).

[2029] Input: Text data, metadata

[2030] Output: Saved database entries

[2031] Specific operation: Connect to the database and save text data and metadata in a specified format.

[2032] Step 3:

[2033] Pre-treatment

[2034] The server retrieves talk papers from the database and performs preprocessing. Specifically, it performs tokenization, stop word removal, and removal of special characters.

[2035] Input: Text data

[2036] Output: Preprocessed text data

[2037] Specific operation: Uses the natural language processing library (nltk) to perform tokenization, stop word removal, and special character removal.

[2038] Step 4:

[2039] Keyword and phrase extraction

[2040] The server extracts important keywords and phrases from the pre-processed text data.

[2041] Input: Preprocessed text data

[2042] Output: List of keywords and phrases

[2043] Specific operation: Use TfidfVectorizer to evaluate and extract keywords and phrases from text based on their importance.

[2044] Step 5:

[2045] Quantifying the discrepancy

[2046] The server uses the TF-IDF algorithm and cosine similarity to calculate the similarity between the talk papers of the high-expert crew and the low-crew, and quantifies the discrepancy.

[2047] Input: List of keywords and phrases

[2048] Output: Quantified deviation data

[2049] Specific operation: Calculates similarity using TfidfVectorizer and cosine_similarity.

[2050] Step 6:

[2051] Emotion analysis

[2052] The server uses an emotion analysis engine to analyze the user's emotions from keywords and phrases within the talk paper.

[2053] Input: List of keywords and phrases

[2054] Output: Sentiment analysis results (positive, negative, neutral, etc.)

[2055] Specific operation: Classifies the sentiment of keywords and phrases using a proprietary sentiment analysis algorithm.

[2056] Step 7:

[2057] Visualization of results

[2058] The server visualizes quantified deviation data and sentiment data and sends the results to the terminal.

[2059] Input: Quantified deviation data, sentiment data

[2060] Output: Visualized data (graphs, charts, etc.)

[2061] Specific operation: Use matplotlib to generate and visualize bar graphs and radar charts.

[2062] Step 8:

[2063] Development and management of training plans

[2064] Users create training plans for their low-level crew members based on visualized results. The server automatically adjusts the training plan based on sentiment data, monitors progress, and provides notifications.

[2065] Input: Visualized data, training plan

[2066] Output: Adjusted training plan, progress notifications

[2067] Specific operation: The server adjusts the care plan based on sentiment data stored in the database, monitors progress, and generates reminders and notifications as needed.

[2068] The specific processing unit 290 transmits the result of the specific processing to the robot 414. In the robot 414, the control unit 46A causes the speaker 240 and the controlled object 443 to output the result of the specific processing. The microphone 238 acquires audio indicating user input for the result of the specific processing. The control unit 46A transmits the audio data indicating user input acquired by the microphone 238 to the data processing unit 12. In the data processing unit 12, the specific processing unit 290 acquires the audio data.

[2069] Data generation model 58 is a type of so-called generative AI (Artificial Intelligence). One example of data generation model 58 is ChatGPT (Internet search<URL: https: / / openai.com / blog / chatgpt> ), Gemini (Internet search) <url: https: gemini.google.com ?hl="ja">Examples of generative AI include the following. The data generation model 58 is obtained by performing deep learning on a neural network. The data generation model 58 is input with prompts containing instructions, and with inference data such as audio data representing speech, text data representing text, and image data representing images. The data generation model 58 infers from the input inference data according to the instructions indicated by the prompts, and outputs the inference results in data formats such as audio data and text data. Here, inference refers to, for example, analysis, classification, prediction, and / or summarization.

[2070] In the above embodiment, an example was given in which specific processing is performed by the data processing device 12, but the technology of this disclosure is not limited thereto, and the specific processing may also be performed by the robot 414.

[2071] Furthermore, the emotion identification model 59, acting as an emotion engine, may determine the user's emotion according to a specific mapping. Specifically, the emotion identification model 59 may determine the user's emotion according to a specific mapping, which is an emotion map (see Figure 9). Similarly, the emotion identification model 59 may also determine the robot's emotion, and the identification processing unit 290 may perform identification processing using the robot's emotion.

[2072] Figure 9 shows an emotion map 400 in which multiple emotions are mapped. In the emotion map 400, emotions are arranged in concentric circles radiating from the center. The closer to the center of the concentric circles, the more primitive the emotions are located. Further out of the concentric circles, emotions representing states and actions arising from mental states are located. Emotion is a concept that includes feelings and mental states. On the left side of the concentric circles, emotions that are generally generated from reactions occurring in the brain are located. On the right side of the concentric circles, emotions that are generally induced by situational judgment are located. Above and below the concentric circles, emotions that are generally generated from reactions occurring in the brain and induced by situational judgment are located. In addition, the emotion of "pleasure" is located on the upper side of the concentric circles, and the emotion of "displeasure" is located on the lower side. Thus, in the emotion map 400, multiple emotions are mapped based on the structure in which emotions arise, and emotions that are likely to occur simultaneously are mapped close together.

[2073] These emotions are distributed at the 3 o'clock position on the Emotion Map 400, and usually fluctuate between feelings of security and anxiety. In the right half of the Emotion Map 400, situational awareness takes precedence over internal feelings, resulting in a calm impression.

[2074] The inside of the Emotion Map 400 represents inner thoughts, while the outside represents actions. Therefore, the further you go from the outside of the Emotion Map 400, the more visible (expressed in actions) your emotions become.

[2075] Here, human emotions are based on various balances, such as posture and blood sugar levels. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. Similarly, in robots, cars, motorcycles, etc., emotions can be created based on various balances, such as posture and battery level. When these balances deviate from the ideal, it results in discomfort, and when they approach the ideal, it results in pleasure. The emotion map can be generated, for example, based on Dr. Mitsuyoshi's emotion map (Research on a system for analyzing brain physiological signals of speech emotion recognition and emotion, Tokushima University, doctoral dissertation: https: / / ci.nii.ac.jp / naid / 500000375379). The left half of the emotion map contains emotions belonging to a region called "response," where sensation is dominant. The right half of the emotion map contains emotions belonging to a region called "situation," where situational awareness is dominant.

[2076] The emotion map defines two emotions that promote learning. One is the emotion around the middle of the negative "repentance" and "reflection" on the situation side. In other words, it is when the robot experiences negative emotions such as "I never want to feel this way again" or "I don't want to be scolded again." The other is the emotion around the positive "desire" on the reaction side. In other words, it is when the robot has positive feelings such as "I want more" or "I want to know more."

[2077] The emotion identification model 59 inputs user input into a pre-trained neural network, obtains emotion values ​​representing each emotion shown in the emotion map 400, and determines the user's emotion. This neural network is pre-trained based on multiple training data sets, which are combinations of user input and emotion values ​​representing each emotion shown in the emotion map 400. Furthermore, this neural network is trained so that emotions located close together have similar values, as shown in the emotion map 900 in Figure 10. Figure 10 shows an example where multiple emotions such as "reassured," "calm," and "confident" have similar emotion values.

[2078] The above description primarily focuses on the functions of the data processing device 12 in relation to this disclosure. However, the system related to this disclosure is not necessarily implemented on a server. The system related to this disclosure may be implemented as a general information processing system. This disclosure may be implemented, for example, as a software program that runs on a personal computer or as an application that runs on a smartphone. The method related to this disclosure may be provided to users in SaaS (Software as a Service) format.

[2079] In the above embodiment, an example was given in which a specific process is performed by a single computer 22. However, the technology of this disclosure is not limited thereto, and a distributed processing of the specific process may be performed by multiple computers, including computer 22. For example, a data generation model 58 may be provided in an external device of the data processing device 12, and the external device may generate data according to the input data.

[2080] In the above embodiment, an example was given in which the specific processing program 56 is stored in the storage 32, but the technology of this disclosure is not limited thereto. For example, the specific processing program 56 may be stored in a portable, computer-readable, non-temporary storage medium such as a USB (Universal Serial Bus) memory. The specific processing program 56 stored in the non-temporary storage medium is installed in the computer 22 of the data processing device 12. The processor 28 executes specific processing according to the specific processing program 56.

[2081] Alternatively, the specific processing program 56 may be stored in a storage device such as a server connected to the data processing device 12 via the network 54, and the specific processing program 56 may be downloaded and installed on the computer 22 in response to a request from the data processing device 12.

[2082] Furthermore, it is not necessary to store the entirety of the specific processing program 56 in a storage device such as a server connected to the data processing device 12 via the network 54, or to store the entirety of the specific processing program 56 in the storage 32; it is acceptable to store only a portion of the specific processing program 56.

[2083] The following types of processors can be used as hardware resources to perform specific processing. Examples of processors include a CPU, a general-purpose processor that functions as a hardware resource to perform specific processing by executing software, i.e., a program. Other examples of processors include dedicated electrical circuits, such as FPGAs (Field-Programmable Gate Arrays), PLDs (Programmable Logic Devices), or ASICs (Application Specific Integrated Circuits), which have circuit configurations specifically designed to perform specific processing. All of these processors have built-in or connected memory, and all of them perform specific processing by using memory.

[2084] The hardware resource that performs a specific process may consist of one of these various processors, or it may consist of a combination of two or more processors of the same or different types (for example, a combination of multiple FPGAs, or a combination of a CPU and an FPGA). Alternatively, the hardware resource that performs a specific process may consist of a single processor.

[2085] Examples of configurations using a single processor include, firstly, a configuration in which one or more CPUs and software are combined to form a single processor, and this processor functions as a hardware resource that performs a specific process. Secondly, there is a configuration using a processor that realizes the functions of the entire system, including multiple hardware resources that perform a specific process, on a single IC chip, as exemplified by SoCs (System-on-a-chip). In this way, a specific process is realized using one or more of the above types of processors as hardware resources.

[2086] Furthermore, the hardware structure of these various processors can more specifically utilize electrical circuits that combine circuit elements such as semiconductor devices. Also, the specific processing described above is merely an example. Therefore, it goes without saying that unnecessary steps can be deleted, new steps added, or the processing order rearranged, as long as it does not deviate from the main purpose.

[2087] The descriptions and illustrations presented above are detailed explanations of the technical aspects of this disclosure and are merely examples of the technical aspects. For example, the above descriptions of the structure, function, operation, and effect are examples of the structure, function, operation, and effect of the technical aspects of this disclosure. Therefore, it goes without saying that you may delete unnecessary parts, add new elements, or replace elements in the descriptions and illustrations presented above, as long as you do not deviate from the essence of the technical aspects of this disclosure. Furthermore, in order to avoid confusion and facilitate understanding of the technical aspects of this disclosure, explanations of common technical knowledge and the like that do not require special explanation to enable the implementation of the technical aspects of this disclosure have been omitted from the descriptions and illustrations presented above.

[2088] All documents, patent applications, and technical standards described herein are incorporated by reference to the same extent as if each individual document, patent application, and technical standard were specifically and individually noted to be incorporated by reference.

[2089] The following is further disclosed regarding the embodiments described above.

[2090] Understood. I propose the following draft claims.

[2091] (Claim 1)

[2092] A means of inputting talk papers for high-expert crew and low-crew,

[2093] A means for saving and pre-processing the input talk paper,

[2094] A means for extracting keywords and phrases from a pre-processed talk paper,

[2095] A means of comparing the frequency and context of extracted keywords and phrases and quantifying the discrepancies,

[2096] A means of visualizing quantified deviation data and displaying the results,

[2097] A system that includes means for formulating and managing training plans based on the displayed results.

[2098] (Claim 2)

[2099] The system according to claim 1, which extracts keywords and phrases from a talk paper using a natural language processing algorithm.

[2100] (Claim 3)

[2101] The system according to claim 1, which uses TF-IDF and cosine similarity algorithms to quantify the degree of deviation.

[2102] "Example 1"

[2103] (Claim 1)

[2104] A means for users to log in and enter data for high-expert crew and low-crew,

[2105] A means of obtaining the data entered by the terminal and sending it to the server,

[2106] A means of storing the data received by the server in a database and recording metadata,

[2107] The server performs preprocessing on the stored data, such as tokenization and stop word removal,

[2108] A server provides means for extracting keywords and important phrases from pre-processed data,

[2109] A means of comparing the frequency and usage of extracted keywords and phrases and quantifying the discrepancies,

[2110] A means by which the server visualizes quantified deviation data, creates graphs and charts, and sends them to the terminal,

[2111] A means for users to formulate training plans based on visualized results and save them on a server,

[2112] The server manages the progress of the training plan and provides a means to offer reminders and notifications.

[2113] A system that includes this.

[2114] (Claim 2)

[2115] The system according to claim 1, which extracts keywords and important phrases from preprocessed data using a natural language processing algorithm.

[2116] (Claim 3)

[2117] The system according to claim 1, which uses TF-IDF and cosine similarity algorithms to quantify the degree of deviation.

[2118] "Application Example 1"

[2119] (Claim 1)

[2120] A means of inputting talk papers for high-expert crew and low-crew,

[2121] A means for saving and pre-processing the input talk paper,

[2122] A means for extracting keywords and phrases from a pre-processed talk paper,

[2123] A means of comparing the frequency and context of extracted keywords and phrases and quantifying the discrepancies,

[2124] A means of visualizing quantified deviation data and displaying the results,

[2125] Based on the displayed results, a means of formulating and managing a training plan,

[2126] A system that includes means for displaying analysis results and is linked to a robot that optimizes operations.

[2127] (Claim 2)

[2128] The system according to claim 1, which extracts keywords and phrases from a talk paper using a natural language processing algorithm.

[2129] (Claim 3)

[2130] The system according to claim 1, which uses TF-IDF and cosine similarity algorithms to quantify the degree of deviation.

[2131] "Example 2 of combining an emotion engine"

[2132] (Claim 1)

[2133] A means of entering text documents for high-expert crews and low-crews,

[2134] Means for saving and pre-processing an input text document,

[2135] A means for extracting keywords and phrases from a preprocessed text document,

[2136] A means of comparing the frequency and context of extracted keywords and phrases and quantifying the discrepancies,

[2137] A means of analyzing emotions from keywords and phrases and saving the results,

[2138] A means for visualizing quantified deviation data and sentiment data, and displaying the results,

[2139] A system that includes means for formulating and managing training plans based on the displayed results.

[2140] (Claim 2)

[2141] The system according to claim 1, which extracts keywords and phrases from a text document using a natural language processing algorithm.

[2142] (Claim 3)

[2143] The system according to claim 1, which uses TF-IDF and cosine similarity algorithms to quantify the degree of deviation.

[2144] "Application example 2 when combining with an emotional engine"

[2145] (Claim 1)

[2146] A means of inputting talk papers for high-expert crew and low-crew,

[2147] A means for saving and pre-processing the input talk paper,

[2148] A means for extracting keywords and phrases from a pre-processed talk paper,

[2149] A means of comparing the frequency and context of extracted keywords and phrases and quantifying the discrepancies,

[2150] A means of visualizing quantified deviation data and displaying the results,

[2151] Based on the displayed results, a means is provided to plan and manage training plans, and to periodically monitor and notify progress.

[2152] A system that includes a means of analyzing user emotions from keywords and phrases within a talk paper using an emotion analysis engine.

[2153] (Claim 2)

[2154] The system according to claim 1, which extracts keywords and phrases from a talk paper using a natural language processin...

Claims

1. A means of inputting talk papers for high-expert crew and low-crew, A means for saving and pre-processing the input talk paper, A means for extracting keywords and phrases from a pre-processed talk paper, A means of comparing the frequency and context of extracted keywords and phrases and quantifying the discrepancies, A means of visualizing quantified deviation data and displaying the results, A system that includes means for formulating and managing training plans based on the displayed results.

2. The system according to claim 1, which extracts keywords and phrases from a talk paper using a natural language processing algorithm.

3. The system according to claim 1, which uses TF-IDF and cosine similarity algorithms to quantify the degree of deviation.

Images