Business support document creation device

Abstract

To provide a business support document creation device for creating business support documents including daily reports with high accuracy.SOLUTION: A business support document creation device 1 includes: an operation log classification unit 112 that classifies operation logs of monitoring object terminals used for business by comparing them with business elements being the business classification indicators for every time frame of a prescribed time; and a document creation unit 113 that creates business support documents including daily reports on the basis of classification results.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a business support document creation device. [Background technology]

[0002] Conventionally, technologies have been developed for analyzing work content and reporting the analysis results. For example, Patent Document 1 discloses a technology that acquires the behavioral history of a user who is a new employee working remotely, and detects at least one of the following specific states of the user: overwork, being stuck, or slacking, using at least one of the working hours information, work content information, and behavioral history information indicated in the acquired behavioral history. Then, a document indicating the status of the new employee is notified to other users who are senior employees.

[0003] As a result, according to the technology of Patent Document 1, other users who are senior employees can grasp the status of the user who is a new employee and take some action such as giving advice. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Publication No. 2022-83228 Summary of the Invention [Problem to be solved by the invention]

[0005] Incidentally, the technology of Patent Document 1 acquires a user's behavioral history and notifies the user of a specific state via a document using at least one of the working hours information, work content information, and behavioral history information indicated in the acquired behavioral history. However, since the detection accuracy is insufficient, there is room for further improvement in terms of the accuracy of the document.

[0006] The present invention has been made in consideration of the above circumstances, and its purpose is to provide a business support document creation device that can create business support documents, such as information indicating a user's specific status, with high accuracy. [Means for solving the problem]

[0007] As a result of intensive research into solving the above-mentioned problems, the inventors have discovered that the above-mentioned object can be achieved by acquiring and classifying operation logs of monitored terminals used for business purposes. As a result, the inventors have completed the present invention. Specifically, the present invention provides the following.

[0008] The present invention includes: an operation log acquisition unit that acquires an operation log of a monitored terminal used for business; an operation log classification unit that classifies the operation logs acquired by the operation log acquisition unit by comparing them with task elements that serve as classification indicators for the tasks for each predetermined time frame; and a document creation unit that creates a business support document including a daily report based on the classification results obtained by the operation log classification unit.

[0009] According to the present invention, it is possible to create business support documents including daily reports with high accuracy based on the classification results obtained by comparing the operation log with business elements for each predetermined time frame. [Effects of the Invention]

[0010] According to the present invention, it is possible to create business support documents including daily reports with high accuracy based on the classification results of the operation logs. [Brief explanation of the drawings]

[0011] [Figure 1] FIG. 1 is an explanatory diagram showing the flow of information in a business support document creation device according to this embodiment. [Figure 2] FIG. 2 is an explanatory diagram showing the flow of information in the business support document creation device of this embodiment. [Figure 3]FIG. 3 is a block diagram showing an example of the hardware configuration of the business support document creation device of this embodiment. [Figure 4] FIG. 4 is a block diagram showing an example of the hardware configuration of the business support document creation device of this embodiment. [Figure 5] FIG. 5 is an explanatory diagram showing a display screen of the terminal to be monitored. [Figure 6] FIG. 6 is an explanatory diagram of the classification result database. [Figure 7] FIG. 7 is a flowchart of the business support document creation program. [Figure 8] FIG. 8 is a flowchart of the business support document creation program. DETAILED DESCRIPTION OF THE INVENTION

[0012] An example of an embodiment of the present invention will be described in detail below with reference to the drawings.

[0013] (Business support document creation device 1) As shown in FIG. 1, the business support document creation device 1 has an operation log acquisition unit 111 that acquires and classifies the operation logs of the monitored terminal 2 used for business, and creates business support documents including daily reports based on the classification results, an operation log classification unit 112 that classifies the operation logs, and a document creation unit 113 that creates business support documents including daily reports based on the classification results.

[0014] (Business support document creation device 1: operation log acquisition unit 111) More specifically, the business support document creation device 1 has an operation log acquisition unit 111 that acquires the operation log of the monitored terminal 2 used in business.

[0015] Here, "work" refers to the various work activities and tasks performed daily by employees who are operators 4 of the monitored terminals 2, and generally includes a series of processes and procedures required by a company or organization to achieve its goals. For example, work includes all forms of work within an organization, such as development, sales, marketing, customer service, and human resources management. The "monitored terminals 2" are information processing devices used within the organization, such as desktop computers, laptop computers, and mobile devices.

[0016] "Operation logs" are comprehensively collected data that records various activities that occur on the monitored terminal 2, and include detailed information on the operation of applications and systems. Operation logs are categorized into application execution logs and system logs. "Application execution logs" include application window titles, file paths (PATH), URLs (Uniform Resource Locators) of accessed websites, the time spent on those windows and pages, the number of typing and mouse clicks, and the capture of image information and collection of on-screen text (including Optical Character Recognition: OCR). These application execution logs are primarily used to analyze the work content and operation patterns of the operator 4 of the monitored terminal 2.

[0017] "System logs" include information such as GPS (Global Positioning System) location information, calendar activity, power on / off, system logon / logoff, sleep mode entry / exit, application launch / exit, printer usage, external device connection / disconnection, file access, creation, deletion, and renaming, etc. These system logs are primarily used to monitor the usage and security of the monitored device 2, and to detect risks of unauthorized access and data leakage.

[0018] (Business support document creation device 1: operation log classification unit 112) The business support document creation device 1 has an operation log classification unit 112 that classifies the operation logs acquired by the operation log acquisition unit 111 by comparing them with business elements that serve as classification indicators for the business for each specified time frame.

[0019] Here, the "predetermined time frame" refers to the time unit used by the business support document creation device 1 to analyze and classify the operation logs. Specifically, it can be one minute, five minutes, ten minutes, thirty minutes, or sixty minutes. It is preferable that the time frame be selectable based on the business content and the purpose of analysis. For example, if a detailed analysis is required, a one-minute or ten-minute time frame can be selected, while a 30-minute or one-hour time frame can be selected to grasp the overall trend. This allows the number of operation logs collected in one day to be reduced to the number of analysis results per time frame, thereby reducing the load of information processing based on the analysis results. The time frame may be, for example, one day, one week, or one month. In this case, if a one-day time frame is used, a daily report can be created for each time frame based on all the operation logs. If a one-week time frame is used, a weekly report can be created, and if a one-month time frame is used, a monthly report can be created. That is, the operation log classification unit 112 can reduce the load of information processing and set the timing for creating daily, weekly, or monthly reports simply by changing the time frame settings.

[0020] "Work elements that serve as work classification indicators" are items that serve as criteria for classifying and analyzing work content. Specifically, examples of work elements include "categories," which are the types of work (e.g., sales, development, clerical work); "customers," which are customer information (e.g., customer name, industry, and sales revenue); "projects," which are project information (e.g., project name, period, and person in charge); "projects," which are project information (e.g., case name, content, and person in charge); "products and services," which are product and service information (e.g., product name, service name, and price); "channels," which are business channels (e.g., telephone, email, and face-to-face); "processes," which are business processes (e.g., order processing, customer support, and invoice issuance); "time," which is work time (e.g., start time, end time, and working hours); and "location," which is the work location (e.g., office, home, or business trip destination).

[0021] "Matching and categorizing" refers to extracting useful information from collected operation logs and organizing it based on specific categories or criteria. Examples of matching and categorizing processes include keyword processing, machine learning processing, and large-scale language processing.

[0022] It is preferable that the operation log classification unit 112 performs a data cleansing process on the collected operation logs before matching. The data cleansing process is a process for effectively extracting and classifying useful information from the operation logs, ensuring the quality of the log data of the operation logs and minimizing the effects of misclassification and noise. For example, it is possible to remove meaningless entries and data with incorrect formats from the log data and organize it into a consistent format.

[0023] Keyword processing is a technique for extracting relevant information from operation logs based on specific keywords or phrases and categorizing it. Specifically, a keyword definition process is performed to predefine keywords and phrases for classification. For example, when classifying sales-related logs, keywords are selected based on terms specific to each business category or words that generally describe the business content, such as "sales," "customer," and "order." Next, the collected operation logs are determined to determine whether they contain predefined keywords, and if an operation log containing a keyword is identified, the operation log is assigned to the business category corresponding to the keyword. For example, an operation log for which "sales" is identified as the keyword will be classified and organized in the "sales" category.

[0024] Machine learning processing is a technique that uses algorithms to learn patterns from operation logs and automatically classify or predict data based on that knowledge. Specifically, a large amount of operation logs and corresponding business elements are collected, and this data is preprocessed to prepare it in a format that is easy for the machine learning model to learn. After that, a machine learning model built with an algorithm such as a neural network is trained using supervised learning. The trained machine learning model learns the relationship between operation logs and business elements, and this model is used to evaluate and adjust the classification accuracy of new operation logs. After passing the evaluation, it is used to classify new operation logs. Furthermore, classification using a machine learning model is more flexible than keyword processing in that it can adapt to new business content, reducing the risk of misclassification and enabling more effective business classification.

[0025] Large-scale language model processing is a method of solving text-based tasks by learning from a wide range of language data and applying contextual understanding and generation capabilities. Specifically, a trained large-scale language model learns the contextual nuances and meanings of language from large amounts of text data, analyzes the content of action logs, matches them with related business elements (e.g., project management, customer support, internal meetings, etc.), and classifies each log entry, i.e., information such as the date and time the event occurred, the type of event, and details of the event, into the appropriate category.

[0026] (Business support document creation device 1: document creation unit 113) The business support document creation device 1 has a document creation unit 113 that creates business support documents including daily reports based on the classification results obtained by the operation log classification unit 112.

[0027] Here, the "classification results" refer to the results of classification based on business elements, such as customer service, technology development, meetings, and document creation. A "daily report" is a document used by employees or teams to record their work activities for the day. It includes the details of the tasks performed that day, the results achieved, any problems encountered, any outstanding issues, and plans for the next work day. Daily reports are an important tool for managers to track the daily progress of their employees (operators 4) and understand the progress of processes and projects. They also promote communication between teams and increase operational transparency. Daily reports enable managers to make appropriate management decisions, such as improving work efficiency, appropriately allocating resources, and implementing prompt countermeasures when necessary. Employees can also use them as a reference for reflecting on their own work and planning their next work.

[0028] "Business support documents" are documents intended to support and improve various business processes within an organization. Business support documents include a wide range of documents, including daily reports, management advice, personnel evaluations, stress diagnoses, labor diagnoses, slacking diagnoses, internal manuals, and documents that visualize work quality. The purpose of business support documents is to organize information within an organization and provide useful insights to managers and employees. For example, daily reports record the day's work activities, achievements, and problems encountered, which allows for continuous monitoring and improvement of operations. Furthermore, management advice and personnel evaluation documents provide information useful for long-term strategy formulation and employee performance evaluation, enabling management and managers to make more effective decisions.

[0029] Furthermore, labor and stress diagnostic documents focus on the quality of the working environment and employee well-being, allowing for concrete proposals to be made to improve workplace health and productivity. Slacking diagnostic documents allow for the identification of incorrect work practices and the implementation of countermeasures. Internal manuals and work quality visualization documents promote the standardization and efficiency of work processes and help employees perform their tasks in a more consistent manner.

[0030] In addition, the business support document creation device 1 may be configured to create business support documents including daily reports by acquiring and classifying the operation logs of the monitored terminal 2 used for business and operating a large-scale language model based on a prompt including the classification result, so that the document creation unit 113 creates business support documents including daily reports using a large-scale language model based on the classification result.

[0031] Specifically, the document creation unit 113 has a large-scale language model that probabilistically predicts how likely words and sentences given in a prompt are to occur in natural language by combining one or more types of processing in natural language processing, such as morphological analysis, syntactic analysis, semantic analysis, contextual analysis, and intention analysis, and is configured to analyze a prompt that instructs the creation of a business support document based on the classification result using the large-scale language model, and to predict and generate a business support document based on the content of the analyzed prompt using the large-scale language model.

[0032] The timing at which the document creation unit 113 creates business support documents is preferably set and managed in the schedule unit 1152 of FIG. 3. This is because the creation timing can be set and changed depending on the type of business support document simply by changing the management data in the schedule unit 1152. Examples of the timing at which business support documents are created include periodic creation of business support documents such as weekly reports and monthly reports, creation of business support documents summarizing monthly business results and issues, creation of business support documents summarizing quarterly business results and issues, and creation of business support documents summarizing annual business results and issues. Other examples of the timing at which business support documents are created include completion of work, such as when a project, an item, or a task is completed, and timing of specific events, such as personnel evaluations, before a management meeting, or before a meeting with a client.

[0033] It is preferable that the processing function of the document creation unit 113 be divided into a function of creating a business support document using a large-scale language model and a function of creating a prompt and outputting the prompt to the large-scale language model. That is, it is preferable that the document creation unit 113 has a large-scale language model unit 1131 having a large-scale language model, and a document creation prompt generation unit 1132 that outputs a prompt to the large-scale language model unit 1131, the prompt having content instructing the creation of a business support document as a document based on the classification result. In this case, the large-scale language model unit 1131 can be used both for the document creation prompt generation unit 1132 and for the other prompt generation units.

[0034] (Business support document creation device 1: document creation unit 113: large-scale language model unit 1131) The document creation unit 113 has a large-scale language model unit 1131 that has a large-scale language model that combines one or more types of processing in natural language processing, such as morphological analysis, syntactic analysis, semantic analysis, contextual analysis, and intention analysis, to probabilistically predict how likely the words and sentences given in the prompt are to occur in natural language, analyze the prompt, and predict and generate documents based on the content of the analyzed prompt.

[0035] Here, "natural language processing" refers to a process that enables a computer to understand text and speech data written in natural language and execute processing appropriate to the purpose. Specifically, examples include morphological analysis, which breaks natural language down into "morphemes," the smallest units that make up the language, and assigns information such as parts of speech; syntactic analysis, which analyzes the grammatical structure of natural language to clarify the structure and meaning of a sentence; semantic analysis, which analyzes the meaning of natural language to understand the meaning of words and sentences and make logical judgments and inferences; contextual analysis, which understands natural language while taking into account the context before and after a sentence; and intent analysis, which extracts the intention of a speaker or writer from a conversation or sentence using natural language. Thus, "natural language processing" processes natural language by combining processes such as morphological analysis, syntactic analysis, semantic analysis, contextual analysis, and intent analysis, and enables the generation of business support documents that support the order-taking activities of this embodiment, as well as machine translation, automatic summarization, question-answering systems, and speech recognition.

[0036] A "prompt" is a word or sentence input to the large-scale language model unit 1131, and serves as the starting point for the large-scale language model to generate business support documents. The generation of business support documents using a large-scale language model based on prompts is a major difference from regular (conventional) machine learning. In regular machine learning, a model learns from training data and predicts output data for input data. For example, a machine learning model for image recognition learns from training data of images of cats and dogs and classifies the input image as either a cat or a dog. On the other hand, in generating business support documents using a large-scale language model, the model learns not only from training data but also from prompts that provide instructions and information regarding the output data that the model is desired to generate.

[0037] Specifically, while conventional machine learning models can only generate content contained in the training data, large-scale language models can use prompts to generate new content not contained in the training data. For example, even if the training data only contains content about female office workers in their 20s, prompts can also generate content about male office workers in their 50s. Furthermore, while conventional machine learning models can only generate variations of content contained in the training data, large-scale language models can use prompts to generate new variations of output data not contained in the training data. For example, even if the training data only contains content from the employee's perspective, prompts can also generate content from the manager's perspective. Furthermore, conventional machine learning models can only generate new content by recursively combining content contained in the training data, while large-scale language models can use prompts to generate creative content not contained in the training data. For example, even if the training data only contains explanations of existing problem-solving methods, prompts can generate problem-solving ideas from completely new perspectives.

[0038] A "large-scale language model" is a type of probabilistic model used in natural language processing, which is a model for probabilistically predicting how likely a given word or sentence is to occur in natural language. Specifically, a language model calculates the occurrence probability of a given word sequence or sentence, or compares the occurrence probabilities of multiple word sequences or sentences, making it possible to automatically generate the most likely word or sentence based on the context when predicting the next word or sentence, or to generate a sentence that meets specific conditions.

[0039] (Business support document creation device 1: document creation unit 113: document creation prompt generation unit 1132) The document creation unit 113 also includes a document creation prompt generation unit 1132 that outputs a prompt to the large-scale language model unit 1131, the prompt instructing the creation of a business support document as a document based on the classification result.

[0040] As described above, it is preferable that the document creation prompt generation unit 1132 exists as a module independent from the large-scale language model unit 1131. The reason for this is that the document creation prompt generation unit 1132 has the role of generating appropriate prompts based on the classification results from the operator 4, which is different from the natural language processing role of the large-scale language model unit 1131. For this reason, by treating the two as independent modules, it is possible to clearly distinguish their respective roles and facilitate system design and maintenance. Furthermore, by implementing the document creation prompt generation unit 1132 as an independent module, it is possible to flexibly respond to changes in the types of business support documents or updates to the large-scale language model.

[0041] The document creation prompt generation unit 1132 has an interface for accepting the classification results received from the action log classification unit 112, and prompt generation logic for generating a prompt based on the received classification results, the prompt instructing the user to create a business support document as a document based on the classification results. The prompt generation logic is used to select a prompt template appropriate for the business support document. Conditional branching is used for this selection, and a prompt template that matches the business support document is selected. The document creation prompt generation unit 1132 also has a prompt transmission interface for transmitting the generated prompt to the large-scale language model unit 1131.

[0042] Prompt templates are prepared in advance. Prompt templates provide the document structure and content framework that form the basis for the effective creation of business support documents in the document creation prompt generation unit 1132. These templates are customized and pre-designed according to the specific needs and business requirements of the organization. Examples include a daily report template for summarizing information about an employee's activities for the day, tasks accomplished, challenges encountered, and plans for the next day; a weekly / monthly report template for providing information on weekly or monthly project progress, team achievements, important events, and goals for the next period; a project status report template for detailed documents containing progress, budget usage, risk assessment, and required action items related to a specific project; a personnel evaluation report template that includes an employee's performance evaluation, strengths and areas needing improvement, and suggestions for career development; and a labor diagnosis report template for analyzing areas of interest to the human resources department, such as an employee's working conditions, working hours, vacation usage, and stress level assessment. The prompt generation logic selects the most appropriate template from these templates and generates specific instructions necessary for document creation. This enables the generation of high-quality business support documents using large-scale language models.

[0043] (Business support document creation device 1: Overall configuration) As described above, the business support document creation device 1 has an operation log acquisition unit 111 that acquires the operation logs of the monitored terminal 2 used for business, an operation log classification unit 112 that classifies the operation logs acquired by the operation log acquisition unit 111 by comparing them with business elements that serve as classification indicators for business for each specified time frame, and a document creation unit 113 that creates business support documents including daily reports based on the classification results obtained by the operation log classification unit 112.

[0044] According to the above configuration, it is possible to create business support documents including daily reports with high accuracy based on the classification results obtained by comparing the operation log with business elements for each predetermined time frame.

[0045] The business support document creation device 1 also has an operation log acquisition unit 111 that acquires operation logs of the monitored terminal 2 used for business operations, an operation log classification unit 112 that compares the operation logs acquired by the operation log acquisition unit 111 with business elements that serve as classification indicators for business operations for each specified time frame and classifies them, and a document creation unit 113 that creates business support documents including daily reports based on the classification results obtained by the operation log classification unit 112.The document creation unit 113 has a large-scale language model unit 1131 that has a large-scale language model that combines one or more types of processing in natural language processing, such as morphological analysis, syntactic analysis, semantic analysis, context analysis, and intention analysis, to probabilistically predict how likely words and sentences given in a prompt are to occur in natural language, analyze the prompt, and predict and generate a document based on the content of the analyzed prompt, and a document creation prompt generation unit 1132 that outputs a prompt to the large-scale language model unit 1131, the content of which instructs the creation of a business support document as a document based on the classification results.

[0046] According to the above configuration, the operation log is compared with business elements for each specified time frame, and the classification results are included in a prompt, and a large-scale language model is operated based on this prompt, making it possible to create business support documents including daily reports with high accuracy.

[0047] The document creation unit 113 may be configured to integrate the large-scale language model unit 1131 and the document creation prompt generation unit 1132. Specifically, the document creation unit 113 may have a large-scale language model that probabilistically predicts how likely words and sentences given in a prompt are to occur in natural language by combining one or more types of processing in natural language processing, such as morphological analysis, syntactic analysis, semantic analysis, context analysis, and intention analysis, analyze a prompt with content instructing the creation of a business support document based on the classification result using the large-scale language model, and predict and generate a business support document based on the content of the analyzed prompt using the large-scale language model.

[0048] The business support document creation device 1 may be configured to provide business support documents to others through an API (Application Programming Interface). When the business support document creation device 1 is configured to provide business support documents to others through an API, it is effective in terms of information accessibility, integration, customization, automation, and scalability. In other words, using an API enables customization according to the specific needs of each department or stakeholder, and the content and format of the document can be flexibly adjusted.

[0049] In the business support document creation device 1, it is preferable that the large-scale language model is specialized for business support. In this case, the large-scale language model unit 1131 has a large-scale language model specialized for creating business support documents using natural language processing based on the business support-related information in the business support-related database 121, thereby improving generation accuracy and efficiency. Specifically, by focusing on learning vocabulary, grammar, example sentences, templates, etc. related to the business support-related information, it is possible to generate business support documents with higher accuracy, and by optimizing the generation process, it is possible to generate business support documents in a shorter time.

[0050] Specifically, the business support document creation device 1 has a business support related database 121 that stores business support related information related to business elements, and the large-scale language model of the large-scale language model unit 1131 is specialized for creating business support documents by learning and adjusting the large-scale language model so as to generate business support documents by sentence generation and question answering based on the business support related information in the business support related database 121. Details of the business support related database 121 will be described later.

[0051] Here, "business support related information" refers to information containing data and knowledge necessary to understand the daily operations of an organization and to improve and streamline them. The business support related information includes employee information on the operators 4 who operate the monitored terminals 2, classification result information classified by the operation log classification unit 112, and business pattern information indicating business patterns per certain period of time, as well as information obtained through operation logs. Examples of information obtained through operation logs include application usage, website visit history, file access and operations, and system usage. This information can clarify what tasks employees are performing, which applications are frequently used, and the times and business patterns at which work efficiency is highest.

[0052] In addition, business support-related information includes activity data such as employee PC usage time, keyboard inputs, and mouse clicks, communication data (emails, chats, meeting notes), project management data (progress, milestone achievement, resource utilization), customer interaction data (inquiries, service requests, customer satisfaction survey results, purchasing history), and financial data (sales, profits, expenses, budget compliance).

[0053] Furthermore, according to the above configuration, as shown in FIG. 2, in the business support document creation device 1, the large-scale language model unit 1131 learns using business support related information such as operator information and operation log information in the business support related database 121 having each database 1211 to 1216, the classification results of the operation log classification unit 112 correspond to explanatory variables of machine learning, and in response to a prompt from the document creation prompt generation unit 1132, the large-scale language model unit 1131 generates business support documents as equivalent to target variables of machine learning.

[0054] As a result, the business support document creation device 1 receives the classification results and generates business support documents using a large-scale language model specialized for creating business support documents, thereby making it possible to fully support business management. In particular, by using a large-scale language model to generate business support documents rather than ordinary machine learning, it is possible to generate new content and variations not included in the training data, making it possible to generate flexible and diverse business support documents according to the task content and situation, as well as to generate creative content not included in the training data.

[0055] Furthermore, because the accuracy of large-scale language models can be improved through training, it is possible to learn from past successes and continuously generate more effective business support documents. In other words, it is possible to analyze the results of business management and generate business support documents that reflect areas for improvement, or to reflect feedback from managers and generate business support documents that are even more satisfying for managers.

[0056] In this way, the business support document creation device 1 uses a large-scale language model to provide advanced functions that are difficult to achieve with conventional machine learning approaches. Specifically, it utilizes the unique capabilities of large-scale language models, such as complex text generation using natural language processing and context-based content generation. Conventional machine learning models typically perform prediction, classification, clustering, and other tasks using primarily numerical and categorical data, and are limited in their ability to process large amounts of text data and generate new text. On the other hand, large-scale language models are adept at learning language patterns from large amounts of text data and generating new text based on given prompts. Therefore, in situations that require the use of diverse and complex language, such as business support documents, an approach using a large-scale language model is more appropriate, and it is possible to provide functions that are difficult to achieve with conventional machine learning methods alone.

[0057] As shown in FIG. 3, the business support document creation device 1 includes an operation log acquisition unit 111, an operation log classification unit 112, and a large-scale language model unit 1131, as well as a terminal control unit 1151, an unacquired log detection unit 1161, an operation log supplementation unit 1162, and a communication unit 13. Furthermore, as shown in FIG. 4, the business support document creation device 1 includes a voice prompt information generation unit 1171, a voice prompt information output unit 1172, a task pattern formation unit 1181, a task change detection unit 1182, and a warning information output unit 1183. Each of the units 111 to 113, 1151, 1161, 1162, 1171, 1172, 1181, 1182, and 1183, except for the communication unit 13, is included in the control unit 11, which is a computer. Some or all of the units included in the control unit 11 may be configured as either hardware or software. Input to the operation log acquisition unit 111 is performed via an input unit 22 such as a keyboard of the monitored terminal 2 operated by the operator 4. In addition, the business support document created by the large-scale language model unit 1131 can be displayed on an output unit 21 such as a display of the monitored terminal 2, and can also be displayed on the output unit of the administrator terminal 7.

[0058] The communication unit 13 is connected to the monitored terminal 2 operated by the operator 4 and the administrator terminal 7 operated by the administrator via the Internet 5 so as to enable data communication. Note that data communication between the communication unit 13 and the monitored terminal 2 and the administrator terminal 7 is not limited to the Internet 5, and may use an information communication network such as a dedicated line or a LAN (Local Area Network), or may use Bluetooth (registered trademark), a wireless communication standard for short-range communication. Data communication between the communication unit 13 and the monitored terminal 2 and the administrator terminal 7 may also be via a dedicated line to prevent information leakage. Furthermore, when data communication between the communication unit 13 and the monitored terminal 2 and the administrator terminal 7 is performed via the Internet 5, it is preferable that the communication unit 13, the monitored terminal 2, and the administrator terminal 7 have a VPN (Virtual Private Network) function. If a VPN function is incorporated into the communication unit 13, the monitored terminal 2, and the administrator terminal 7, for example, data communication from the monitored terminal 2 passes through a VPN connection, thereby protecting the communication from unauthorized external access and ensuring secure communication. The monitored terminal 2, the administrator terminal 7, and the administrator terminal 7 are information processing devices such as general personal computers, laptop computers, smartphones, and tablet terminals.

[0059] The terminal control unit 1151 has a user interface function that sets the display screens of the monitored terminal 2 and the manager terminal 7 to screens suitable for viewing business support documents and inputting classification results. Specifically, the terminal control unit 1151 makes it possible to check business support documents, classification results, etc. For example, as shown in FIG. 5, if the business support document is a daily report, the contents of the daily report are displayed on the screen, and comments can be entered in predetermined areas of the daily report on both the monitored terminal 2 and the manager terminal 7. The contents of the daily report include information such as the employee, job title, workplace name, work summary, deliverables, and work evaluation.

[0060] As shown in FIG. 3 , the business support document creation device 1 includes an input device 15 connected to an input receiving unit 1192 and a display device 14 connected to a display control unit 1191. Examples of the input device 15 include a keyboard, a mouse, a touch panel, and a voice input device. Examples of the display device 14 include a liquid crystal display device. This allows the business support document creation device 1 to be configured using an information processing device such as a general personal computer, a laptop computer, a smartphone, or a tablet terminal. The business support document creation device 1 may lack at least one of the input device 15 and the display device 14. In this case, the input device 15 and the display device 14 are provided in an external terminal (not shown), and the business support document creation device 1 can be used as a business support document creation server. Furthermore, the business support document creation device 1 may be used as a monitored terminal 2, allowing the operator 4 to proceed with a task while receiving support information generated by the large-scale language model unit 1131.

[0061] In this embodiment, the case where the functions of the business support document creation device 1 are installed in an information processing device is described, but the present invention is not limited to this, and the functions of the business support document creation device 1 may be installed in cloud computing. In this case, cloud computing allows computer resources to be added as needed, making it possible to process large amounts of operation log information, etc., enabling faster and more efficient processing, and also making it possible to easily expand processing capacity to accommodate a significant increase in the number of monitored terminals 2.

[0062] The business support document creation device 1 also has an unacquired log detection unit 1161 that detects a time frame in which an operation log was not acquired by the operation log acquisition unit 111 before the matching is performed in the operation log classification unit 112, and an operation log supplementation unit 1162 that, when the unacquired log detection unit 1161 detects a time frame in which an operation log was not acquired, accepts input of a comment from the operator 4 operating the monitored terminal 2 and makes it into an operation log for the time frame.

[0063] The unacquired log detection unit 1161 has an acquired log analysis function that analyzes the log data of the operation log acquired by the operation log acquisition unit 111 and checks whether or not there is an operation log in each time frame, and an unacquired log detection function that, if there is no operation log in a predetermined time frame based on the analysis result, detects that time frame as an unacquired log. Note that the unacquired log detection unit 1161 preferably has a detection result output function that outputs information that may be related to the time frame of the detected unacquired log.

[0064] The operation log supplementation unit 1162 has a time frame presentation function that presents to the operator the time frame of the missing logs detected by the missing log detection unit 1161, a comment input form provision function that provides a form for the operator to input comments regarding the time frame of the missing logs, a comment saving function that saves the comments input by the operator, and a comment registration function that registers the saved comments as an operation log for the time frame. As a result, the operation log supplementation unit 1162 not only compensates for missing logs by the comment input by the operator 4, but also makes it possible to grasp the business content and situation for that time frame in more detail. The input comments are an important source of information when creating business support documents, enabling the creation of more accurate documents.

[0065] According to the above configuration, even if an operation log is not obtained within a specified time frame for some reason, the operator 4 of the monitored terminal 2 can enter a comment to create an operation log, so that operation logs exist for all time frames, resulting in business support documents with even higher accuracy.

[0066] As shown in Figure 4, the business support document creation device 1 has a business pattern formation unit 1181 that forms a business pattern for a certain period of time of business performed on the monitored terminal 2 based on the classification results in the operation log classification unit 112, a business change detection unit 1182 that detects a qualitative change in the business performed on the monitored terminal 2 based on the amount of change between the business patterns repeatedly formed at certain periods, and a warning information output unit 1183 that, when a qualitative change in business is detected, outputs the monitored terminal 2 in which the qualitative change has occurred, along with the details of the qualitative change, to the administrator terminal 7 as warning information.

[0067] Here, "work patterns" are models for analyzing and characterizing an employee's daily work behavior in terms of time and content. Work patterns represent an employee's daily routine and weekly work schedule, helping to understand work priorities and efficiency, such as which tasks are considered important and which tasks take up the most time. Work patterns can also reveal unique work activities that occur during specific times or days of the week, such as data entry work concentrated in the first few hours of the morning or meetings scheduled on specific days of the week. Furthermore, changes in work patterns can capture fluctuations in employee behavior. For example, if a work pattern shows increased time spent using project-related applications as a project deadline approaches, this could indicate the progress of the project or increasing pressure on the employee.

[0068] According to the above configuration, the business support document creation device 1 analyzes the work patterns of employees on the monitored terminal 2, detects changes in those patterns, and reports them to the manager, thereby significantly improving the efficiency of the organization's business operations and employee performance management. Specifically, the work activity patterns of employees are identified from the data obtained by the action log classification unit 112, and the work pattern formation unit 1181 periodically analyzes the patterns to form a work flow. This clarifies an overview of which tasks are performed, how often, and at what time.

[0069] Next, the work change detection unit 1182 compares the formed work patterns at regular intervals and monitors any qualitative changes that occur during that time. It then detects whether the work flow, volume, or frequency of work is different from normal, and, based on the detection results, evaluates employee efficiency, stress levels, and signs of potential problems (e.g., fatigue or dissatisfaction). If a qualitative change is discovered, the warning information output unit 1183 notifies the manager terminal 7 of the specific changes and which employees are affected. This allows the business support document creation device 1 to quickly provide appropriate support and intervention to problematic employees and improve the overall work environment by adjusting business processes.

[0070] Furthermore, the business support document creation device 1 has a voice-over information prompt generation unit 1171 that outputs a prompt to the large-scale language model unit 1131 to instruct the large-scale language model unit 1131 to generate voice-over information that is useful to the operator corresponding to the business support document based on the business support document created by the document creation unit 113, and a voice-over information output unit 1172 that outputs the voice-over information generated by the voice-over information output unit 1172 to the monitored terminal 2 operated by the operator 4.

[0071] According to the above configuration, the prompt generator 1171 analyzes the data and analysis results contained in the business support document and generates prompts based on the data and analysis results, including information such as the next action the operator should take and points to note. The generated prompts are then sent to the large-scale language modeler 1131, where they are converted into actual text-format prompt information using natural language processing technology. Once generated, the prompt information is output to the monitored terminal 2 used by the operator 4 via the prompt information output unit 1172. This allows the operator 4 to receive direct feedback and instructions regarding their own work in real time, improving work efficiency and preventing misunderstandings. Furthermore, the time-consuming task of prompting, which was previously performed by a human manager, is automated, and appropriate feedback and instructions are instantly provided to the operator 4. This is particularly effective when quickly addressing problems that require urgent correction or business processes that require streamlining.

[0072] (Business Support Related Database 121) Next, the business support related database 121 will be described in detail. The business support related database 121 is stored in the storage unit 12 as shown in FIG. 2. The storage unit 12 is connected to the control unit 11 so as to be able to communicate data. The storage unit 12 may be configured with a hard disk, or may be configured with a combination of a hard disk and memory. In the case of a configuration in which a hard disk and memory are combined, some of the data and indexes used by the database are cached in the memory as needed, thereby enabling faster access to the database. The storage unit 12 may be a data server connected to an information communication network such as the Internet 5, separate from the business support document creation device 1. The storage unit 12 may also be configured with multiple data servers provided for each database.

[0073] The business support related database 121 includes an employee database 1211 , an action log database 1212 , a classification result database 1213 , a chat log database 1214 , a business pattern database 1215 , and other databases 1216 .

[0074] The employee database 1211 is a database that stores information related to employee behavior, work performance, and behavioral patterns. Specifically, the database stores basic personal information such as the employee's name, position, and department, work history such as projects the employee was involved in, tasks they were responsible for, and results they achieved, work behavior data obtained from operation log analysis such as applications used, websites visited, time spent on tasks, and frequency, performance indicators such as performance evaluation scores, goals achieved, and the quality and quantity of documents submitted, communication data such as records of communication via email and chat tools, and attendance information such as attendance status and records of participation in meetings and training sessions.

[0075] The operation log database 1212 is a database that records operation logs of a series of activities and events that occur on monitored terminals 2, such as employees' computers and mobile terminals. Specifically, it stores information such as application execution logs, system logs, keyboard and mouse usage data, screen captures and OCR data.

[0076] The classification result database 1213 is a database that stores the results of sorting and classifying information obtained from operation log data according to specific categories and indices. Specifically, as shown in Fig. 6, it has items such as start time, caption, operation time (h), total workload, employee name, application, exe name, major category, minor category, workplace name, and job title, and the classification results of the operation log are stored in association with each item, thereby forming one item of classification result information.

[0077] Here, "Start Time" indicates the exact time each task began, allowing for analysis of the chronological flow of work and the allocation of work time. "Caption" indicates the main work activities performed during that time period, providing an overview of the work content. "Working Time" indicates the amount of time allocated to each task, allowing for estimation of the importance and urgency of each task. "Total Work Volume" refers to the number of specific tasks or steps required to complete a task. For example, it is the actual number of activities performed during a specific period, such as typing documents, writing code, sending emails, and responding to inquiries, allowing for quantitative evaluation of the amount of work an employee performed within a specific period. "Application / exe name" indicates the software used by the employee to perform the work, providing information about the nature of the work and the employee's technical skills. "Major Category / Subcategory" indicates broad and specific work classifications. "Work Location" indicates the employee's work environment, including whether they work remotely or in an office. "Job Title" indicates the employee's position.

[0078] The classification result database 1213 in Figure 6 records the work content of employee 1 on October 2, 2023. Based on the information in this classification result database 1213, the business support document creation device 1 can determine the time spent on each task and the total work time from the start and end times. Specifically, it can be seen that employee 1 worked for 12.4 hours on October 2, 2023, and the task that took up the most time was bug fixing, at 6.06 hours. It can also be seen that employee 1 worked on various types of tasks in the document management, quality assurance, and development categories, and that he worked from home. This allows the business support document creation device 1 to analyze how much time employee 1 spent on each task by calculating the average time spent on each task based on the classification result database 1213.

[0079] In addition, the start and end times can be used to analyze which time period Employee 1 works the most. The application names and exe names can be used to analyze which applications Employee 1 uses most frequently. The work location names can be used to analyze how much work Employee 1 does working from home and in the office. Based on these analyses, the business support document creation device 1 can, for example, identify tasks that Employee 1 spends a lot of time on but produces little results. By analyzing the workload for each application, the device can suggest work tools to improve Employee 1's work efficiency. By analyzing the workload for each work location, the device can suggest improvements to the work environment to increase Employee 1's concentration and improve their work efficiency. In other words, the business support document creation device 1 can perform a series of processes consisting of situation understanding, situation analysis, and proposals using the classification result database 1213 and a large-scale language model.

[0080] The chat log database 1214 is a database for collecting and analyzing chat communications between employees within an organization. The chat log database 1214 collects information from chat platforms that employees use on a daily basis and stores that information to enable detailed analysis of communication patterns and content. It records information such as the text content of messages sent and received, the sender and recipient of the message, the time of sending, chat group information, and tags for related projects and tasks. This makes it possible to understand the communication efficiency of individual employees and teams, the level of cooperation, and the liveliness of project-related discussions.

[0081] The work pattern database 1215 is a database for comprehensively understanding patterns of employee behavior and work schedules. It collects and stores information such as the flow of employees' daily work, time allocation, priorities, and work efficiency in an analyzable format. Specifically, it provides data such as how much time employees allocate to each task, which tasks take the most time, and which tasks have the highest priority. It also stores information on the time periods and days of the week when tasks are performed, as well as the periodicity of tasks and behavioral patterns related to specific tasks. This allows the business support document creation device 1 to capture changes based on the work pattern database 1215. For example, it can quantitatively detect changes in employee behavior, such as changes in task distribution according to the progress of a project, fluctuations in stress levels, and increases or decreases in workload.

[0082] The other databases 1216 are changed as appropriate depending on the type of business support document. For example, in the case of generating a business support document specialized in management advice, the other databases 1216 may include a financial database that records financial conditions such as sales, profits, expenses, and cash flow, a market analysis database that stores information such as market trends, competitive analysis, customer purchasing behavior, and market share, a strategic planning database that stores long-term strategic plans, short-term goals, and action plans, a human resources management database that stores employee performance data, skill sets, and leadership evaluations, and an operational efficiency database that stores the operational efficiency of each department and team, the results of process improvement, optimized operational procedures, and the like.

[0083] (Business support document creation program) 3 and 4, in the business support document creation device 1, the operation log acquisition unit 111, operation log classification unit 112, large-scale language model unit 1131, terminal control unit 1151, unacquired log detection unit 1161, operation log supplementation unit 1162, prompt information generation unit 1171, prompt information output unit 1172, task pattern formation unit 1181, task change detection unit 1182, and warning information output unit 1183 may be configured as either hardware or software. Each of these units 111 to 113, 1151, 1161, 1162, 1171, 1172, 1181, 1182, and 1183 is included in the control unit 11, which is a computer. When configured as software, the control unit 11 is configured to cause the computer (control unit 11) to execute a business support document creation program that generates business support documents by utilizing a large-scale language model.

[0084] 7, the business support document creation program is a program for causing a computer (control unit 11) to execute an operation log acquisition processing step (S1) for acquiring operation logs of the monitored terminal 2 used for business, an operation log classification processing step (S2) for classifying the operation logs acquired in the operation log acquisition processing step (S1) by comparing them with business elements that serve as classification indicators for business for each predetermined time frame, and a document creation processing step (S3) for creating business support documents including daily reports as documents based on the classification results. Note that the business support document creation program may be configured to cause the computer to execute the functions of each unit in the control unit 11 as processing steps.

[0085] According to the above configuration, it is possible to create business support documents including daily reports with high accuracy based on the classification results obtained by comparing the operation log with business elements for each predetermined time frame.

[0086] As shown in FIG. 8, the business support document creation program is a program for causing a computer (control unit 11) having a large-scale language model unit 1131 with a large-scale language model that combines one or more types of processing in natural language processing, such as morphological analysis, syntactic analysis, semantic analysis, context analysis, and intention analysis, to probabilistically predict how likely words and sentences given in a prompt are to occur in natural language, analyze the prompt, and predict and generate documents based on the content of the analyzed prompt, to execute the following steps: an operation log acquisition processing step (S11) for acquiring the operation log of the monitored terminal 2 used for business; an operation log classification processing step (S12) for classifying the operation log acquired in the operation log acquisition processing step (S11) by comparing it with business elements that serve as classification indicators for the business for each specified time frame; and a document creation prompt generation processing step (S13) for outputting a prompt to the large-scale language model unit 1131 based on the classification results, the prompt instructing the creation of business support documents including a daily report as a document.

[0087] According to the above configuration, the operation log is compared with business elements for each specified time frame, and the classification results are included in a prompt, and a large-scale language model is operated based on this prompt, making it possible to create business support documents including daily reports with high accuracy.

[0088] Furthermore, simply by installing the business support document creation program in an information processing device such as a personal computer or a tablet terminal, the information processing device can function as the business support document creation device 1. The program may be distributed in a state recorded on a computer-readable recording medium such as a CD-ROM or USB memory, or may be distributed via a two-way communication network or communication line such as the Internet or a one-way communication network such as a television broadcast.

[0089] (How to prepare business support documents) The business support document creation device 1 is configured to cause a computer (controller 11) to execute a business support document creation method. Specifically, the method causes the computer (controller 11) to execute an operation log acquisition process for acquiring operation logs of the monitored terminals 2 used for business, an operation log classification process for classifying the operation logs acquired in the operation log acquisition process by comparing them with business elements that serve as classification indicators for business for each predetermined time frame, and a document creation process for creating business support documents including daily reports as documents based on the classification results.

[0090] In addition, by combining one or more types of processes in natural language processing such as morphological analysis, syntactic analysis, semantic analysis, context analysis, and intention analysis, the computer (control unit 11) has a large-scale language model unit 1131 having a large-scale language model that probabilistically predicts how likely words and sentences given in a prompt are to occur in natural language, analyzes the prompt, and predicts and generates documents based on the content of the analyzed prompt.The method is to have the computer (control unit 11) execute an operation log acquisition process that acquires the operation log of the monitored terminal 2 used for business, an operation log classification process that compares the operation log acquired in the operation log acquisition process with business elements that serve as classification indicators for the business for each specified time frame and classifies it, and a document creation prompt generation process that outputs a prompt to the large-scale language model unit 1131 based on the classification results, the content of which instructs the creation of business support documents including daily reports as documents.

[0091] It should be noted that within the scope of the concept of the present invention, those skilled in the art may conceive of various modifications and alterations. Therefore, it is understood that such modifications and alterations fall within the scope of the present invention. For example, those skilled in the art may appropriately add, delete, or modify components of the above-described embodiments, or may add, omit, or change the conditions of processes, as long as they maintain the essence of the present invention. [Explanation of symbols]

[0092] 1. Business support document creation device 2. Monitored devices 4 Operator 5. Internet 7 Administrator terminal 11 Control section 12 Storage section 111 Operation log acquisition unit 112 Operation log classification unit 113 Document Preparation Department 1131 Large-scale Language Modeling 1132 Document Creation Prompt Generation Unit 1151 Terminal control unit 1152 Schedule Department

Claims

1. an operation log acquisition unit that acquires an operation log of a monitored terminal used for business; an operation log classification unit that classifies the operation logs acquired by the operation log acquisition unit by comparing them with task elements that serve as classification indicators for the tasks for each predetermined time frame; a document creation unit that creates business support documents including daily reports based on the classification results obtained by the operation log classification unit. Business support document creation device.

2. The document creation unit a large-scale language model unit having a large-scale language model that combines one or more types of processing in natural language processing, such as morphological analysis, syntactic analysis, semantic analysis, context analysis, and intention analysis, to probabilistically predict how likely words and sentences given in a prompt are to occur in natural language, analyze the prompt, and predict and generate documents based on the content of the analyzed prompt; a document creation prompt generation unit that outputs the prompt, the content of which instructs the creation of the business support document as the document based on the classification result, to the large-scale language model unit.

2. The business support document creation device according to claim 1.

3. a business support related database in which business support related information relating to the business elements is stored; The large-scale language model is The system is specialized for creating business support documents by learning and adjusting the system to generate the business support documents through sentence generation and question answering based on the business support related information in the business support related database.

3. The business support document creation device according to claim 2.

4. an unacquired log detection unit that detects the time frame during which the operation log was not acquired by the operation log acquisition unit before the execution of the matching by the operation log classification unit; an operation log supplementation unit that, when the time frame in which the operation log has not been acquired is detected by the unacquired log detection unit, accepts input of a comment from an operator who operates the terminal to be monitored and makes the comment the operation log for the time frame.

3. The business support document creation device according to claim 2.

5. a task pattern forming unit that forms a task pattern for a certain period of time of tasks executed on the monitored terminal based on the classification result of the operation log classifying unit; a task change detection unit that detects a qualitative change in the task executed on the monitored terminal based on the amount of change between the task patterns repeatedly formed at the fixed intervals; and an alert information output unit that, when a qualitative change in the business is detected, outputs to an administrator terminal information about the monitored terminal in which the qualitative change has occurred together with the details of the qualitative change as alert information.

3. The business support document creation device according to claim 2.

6. a prompt generation unit for generating prompts for the large-scale language model unit to generate useful prompt information for an operator corresponding to the business support document based on the business support document generated by the document generation unit; a voice message output unit that outputs the voice message generated by the voice message prompt generation unit to the monitored terminal operated by the operator.

3. The business support document creation device according to claim 2.

Images