Information processing systems, information processing methods, and programs

Abstract

To efficiently generate summaries of business opportunity information according to each company's specific summarization criteria. [Solution] The information processing system includes a business negotiation information acquisition unit that acquires business negotiation information showing the content of business negotiations, and a summary condition storage unit that stores summary conditions corresponding to each company. The information processing system also includes a summary processing unit that generates a summary result of business negotiation information by inputting the summary conditions and business negotiation information corresponding to the company related to the business negotiation information to be summarized into a large-scale language model.

Description

Technical Field

[0001] The present invention relates to an information processing system, an information processing method, and a program.

Background Art

[0002] Techniques have been proposed for obtaining text meeting minutes, generating meeting paragraph summaries, and generating meeting minutes based on the meeting summary and meeting instructions (see Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] In the conventional technology, it has been difficult to efficiently generate summaries of negotiation information corresponding to different summarization needs for each company.

[0005] The present invention has been made in view of such a background, and an object thereof is to provide a technology capable of efficiently generating a summary of negotiation information according to the summary conditions for each company.

Means for Solving the Problems

[0006] The main invention of the present invention for solving the above problems is an information processing system, comprising: a negotiation information acquisition unit that acquires negotiation information indicating the content related to a negotiation; a summary condition storage unit that stores, for each company, summary conditions corresponding to the company; and a summary processing unit that inputs the summary conditions corresponding to the company and the negotiation information related to the negotiation information to be summarized into a large language model to generate a summary result of the negotiation information.

[0007] Further issues and solutions disclosed in this application will be made clear in the section on embodiments of the invention and in the drawings. [Effects of the Invention]

[0008] According to the present invention, summaries of business negotiation information can be efficiently generated according to the summarization conditions for each company. [Brief explanation of the drawing]

[0009] [Figure 1] This figure shows an example of the overall configuration of an information processing system. [Figure 2] This figure shows an example of the hardware configuration of management server 2. [Figure 3] This figure shows an example of the software configuration for management server 2. [Figure 4] This diagram illustrates the summarization processing flow of business negotiation information in an information processing system. [Figure 5] This diagram illustrates the update process flow for summary conditions in an information processing system. [Figure 6] This diagram illustrates the processing flow for predicting the progress of business negotiations and generating improvement plans in an information processing system. [Figure 7] This diagram illustrates the adaptive control processing flow in an information processing system. [Modes for carrying out the invention]

[0010] <System Overview> The following describes an information processing system according to one embodiment of the present invention. The information processing system of this embodiment acquires negotiation information indicating the content of a business negotiation and generates a summary result of the negotiation information using a Large Language Model (LLM) based on summarization conditions set for each company. The information processing system of this embodiment can respond to the different summarization needs of each company and can efficiently generate negotiation summaries according to the items and format required by each company. Furthermore, the information processing system of this embodiment can also be equipped with a function to evaluate the quality of the generated summary result and automatically update the summarization conditions based on the evaluation result, as well as a function to predict the progress of the negotiation and generate an improvement plan.

[0011] Figure 1 shows an example of the overall configuration of an information processing system. The information processing system in this embodiment includes a management server 2. The management server 2 is connected to the user terminal 1 via a communication network. The communication network is, for example, the internet and is constructed using public telephone networks, mobile phone networks, wireless communication channels, Ethernet (registered trademark), etc.

[0012] User terminal 1 is a computer operated by the user. User terminal 1 can be, for example, a smartphone, a tablet computer, or a personal computer.

[0013] The management server 2 may be a general-purpose computer such as a workstation or personal computer, or it may be logically implemented through cloud computing.

[0014] <Management Server 2> Figure 2 shows an example of the hardware configuration of the management server 2. Note that the illustrated configuration is just one example, and other configurations are also possible. The management server 2 includes a CPU 201, memory 202, storage device 203, communication interface 204, input device 205, and output device 206. The storage device 203 stores various data and programs, such as a hard disk drive, solid-state drive, or flash memory. The communication interface 204 is an interface for connecting to a communication network, such as an adapter for connecting to Ethernet®, a modem for connecting to a public telephone network, a wireless communication device for wireless communication, or a USB (Universal Serial Bus) connector or RS232C connector for serial communication. The input device 205 is for inputting data, such as a keyboard, mouse, touch panel, button, or microphone. The output device 206 is for outputting data, such as a display, printer, or speaker. Furthermore, each functional unit of the management server 2, as described later, is realized by the CPU 201 reading programs stored in the storage device 203 into memory 202 and executing them, and each storage unit of the management server 2 is realized as part of the storage area provided by memory 202 and storage device 203.

[0015] Figure 3 shows an example of the software configuration of the management server 2. The management server 2 includes a sales opportunity information acquisition unit 211, a summarization processing unit 212, an evaluation unit 213, a progress prediction unit 214, an improvement plan generation unit 215, a continuous monitoring unit 216, an adaptive control unit 217, a summary condition update unit 218, an output unit 219, a summary condition storage unit 231, a user information storage unit 232, an evaluation condition storage unit 233, a sales opportunity history storage unit 234, and a model storage unit 235.

[0016] <Management Server 2> The following describes the functional components of the management server 2.

[0017] The summary condition storage unit 231 stores summary conditions corresponding to each company. The summary conditions are conditions used when generating a summary from negotiation information, and include, for example, conditions regarding summary items to be extracted from the negotiation information, the format of the summary result, the length of the summary result, etc. Examples of summary items include, for example, the purpose of the negotiation, the proposed content, the customer's reaction, the next action, risk items, decision maker information, budget scale, introduction time, etc. The summary condition storage unit 231 can store a different set of summary items for each company. In this embodiment, the "company" to which the summary conditions are associated is the company (the user's own company) to which the user who requests the creation of the summary belongs. That is, the summary condition storage unit 231 stores summary conditions that define the summary items and formats required by each company to which the user belongs. Configurations for storing summary conditions corresponding to the company of the negotiation partner and for storing summary conditions for each combination of the user's own company and the company of the negotiation partner will be described later as modification examples (see Modification Example 13 and Modification Example 14). As a specific example of the summary conditions, for example, as the summary conditions corresponding to a certain company X, the summary items are "purpose of the negotiation", "proposed content", "customer's reaction", "next action", the format of the summary result is a list for each summary item, and the length of the summary result is within 100 characters for each summary item. Also, as the summary conditions corresponding to another company Y, the summary items are "decision maker information", "budget scale", "introduction time", "competitive situation", the format of the summary result is a table format, and the conditions include including an example of the summary result related to the past negotiations of the company Y as exemplary data for few-shot learning. Thus, the summary condition storage unit 231 can store combinations of different summary items, formats, lengths, and exemplary data as summary conditions for each company.

[0018] The summary condition storage unit 231 can store exemplary data for few-shot learning as part of the summary conditions. The exemplary data includes an example of negotiation information and an example of the summary result corresponding to the negotiation information. By using the exemplary data, it is possible to specifically show the large language model what format and content the summary should be generated in. The summary condition storage unit 231 can store a plurality of exemplary data for each company.

[0019] The summary condition storage unit 231 has a summary condition setting function. The summary condition setting function is a function that receives and stores the summary conditions input by the user. The user can input, for example, summary items and example data suitable for the company via the user terminal 1. The summary condition storage unit 231 stores the input summary conditions in association with the company to which the user belongs.

[0020] The user information storage unit 232 stores, for each user, company affiliation information that identifies the company to which the user belongs. The user information storage unit 232 stores the user's identification information in association with the identification information of the company to which the user belongs. The user information storage unit 232 can also store the type of authority of the user. The type of authority is information indicating, for example, a classification such as an administrator or a general user. A user with administrator authority can have the authority to change the summary conditions of the company.

[0021] The evaluation condition storage unit 233 stores evaluation criteria for the summary results. The evaluation criteria are the criteria for evaluating the quality of the generated summary results. The evaluation condition storage unit 233 stores, for example, keywords or key phrases that should be included in the summary results, conditions related to the length of the summary results, conditions related to the comprehensiveness of the summary results, conditions related to the conciseness of the summary results, etc. The evaluation condition storage unit 233 can also store different evaluation criteria for each company.

[0022] The negotiation history storage unit 234 stores information related to past negotiations for each company. The negotiation history storage unit 234 stores negotiation information, summary results, negotiation date and time, participant information, negotiation progress status, conclusion results, etc. The negotiation history storage unit 234 can store information spanning multiple negotiations in chronological order. The information stored in the negotiation history storage unit 234 is used for predicting the negotiation progress status and generating improvement plans.

[0023] The model storage unit 235 stores large-scale language models and machine learning models. The large-scale language models are pre-trained models for natural language processing and can be transformer-based models such as GPT® (Generative Pre-trained Transformer) or BERT (Bidirectional Encoder Representations from Transformers). The model storage unit 235 can store a first large-scale language model for summary generation and a second large-scale language model for evidence explanation generation. The machine learning models are pre-trained models for predicting the progress of business negotiations and can be models such as neural networks, decision trees, random forests, or gradient boosting. Note that the large-scale language models and machine learning models stored in the model storage unit 235 are not limited to those held internally by the management server 2, but may also be provided as an external service. In this case, the model storage unit 235 stores information for using a large-scale language model or machine learning model provided by an external service (e.g., API endpoint, authentication information, etc.), and each functional unit, including the summarization processing unit 212, may use this information to utilize the large-scale language model or machine learning model via an external API.

[0024] The business negotiation information acquisition unit 211 acquires business negotiation information that indicates the content of a business negotiation. The business negotiation information acquisition unit 211 can acquire business negotiation information in various ways. As a first method, the business negotiation information acquisition unit 211 can acquire audio data related to a business negotiation and acquire business negotiation information consisting of text data by performing speech recognition processing on the audio data. The audio data may be, for example, audio recorded at the business negotiation or audio recorded in an online meeting system. The business negotiation information acquisition unit 211 converts the audio data into text data using a speech recognition engine. The speech recognition engine may be built into the management server 2 or may be provided as an external service.

[0025] As a second method, the business negotiation information acquisition unit 211 can receive text data related to business negotiations as input and acquire said text data as business negotiation information. The text data may be, for example, text entered by a user via the user terminal 1, or text registered in the business negotiation management system. The business negotiation information acquisition unit 211 can acquire business negotiation information by receiving text data transmitted from the user terminal 1.

[0026] As a third method, the business negotiation information acquisition unit 211 can cooperate with an external business negotiation management system to acquire text data, participant information, and date information related to business negotiations recorded in the said business negotiation management system. The business negotiation information acquisition unit 211 can acquire data via an API (Application Programming Interface) provided by the business negotiation management system. The business negotiation information acquisition unit 211 stores the acquired text data as business negotiation information and the acquired participant information and date information as information attached to the business negotiation information.

[0027] As a fourth method, the sales negotiation information acquisition unit 211 can acquire audio data and screen sharing data obtained from the online meeting system. The sales negotiation information acquisition unit 211 generates text data from the audio data through speech recognition processing and extracts document title information from the screen sharing data. The sales negotiation information acquisition unit 211 generates sales negotiation information by combining the text data and the document title information. The document title information may include, for example, the file name of the screen-shared document or the title written in the document. Note that the information acquired by the sales negotiation information acquisition unit 211 from the screen sharing data is not limited to document title information. The sales negotiation information acquisition unit 211 may also extract various types of information that supplement the conversation data from the screen sharing data, such as the main text, figures, tables, numerical values, and other information written in the document, and generate sales negotiation information by combining this information with the text data.

[0028] The business negotiation information acquisition unit 211 stores the acquired business negotiation information in the business negotiation history storage unit 234. The business negotiation information acquisition unit 211 can store the business negotiation information along with the business negotiation identification information, the date and time of the business negotiation, participant information, and identification information of companies related to the business negotiation.

[0029] The summarization processing unit 212 inputs summarization conditions and deal information corresponding to companies related to the deal information to be summarized into a large-scale language model to generate a summary result of the deal information. First, the summarization processing unit 212 identifies the companies related to the deal information to be summarized. The summarization processing unit 212 can identify companies by referring to the company identification information attached to the deal information. Alternatively, the summarization processing unit 212 can identify the affiliated company information corresponding to the user associated with the deal information to be summarized from the user information storage unit 232 and identify the company corresponding to the identified affiliated company information.

[0030] The summarization processing unit 212 retrieves the summarization conditions corresponding to the identified company from the summarization condition storage unit 231. The summarization processing unit 212 generates a prompt containing the retrieved summarization conditions and deal information. The prompt is text to be input to the large-scale language model and includes instructions and input data for the large-scale language model. The summarization processing unit 212 includes in the prompt a description instructing the large-scale language model to summarize so that each summary item included in the summarization conditions includes a summary result.

[0031] For example, if the summarization criteria include four summarization items: "Deal Objective," "Proposal," "Customer Response," and "Next Action," the summarization processing unit 212 will include instructions in the prompt such as, "Summarize the following deal information. The summary should include information for each of the following items: Deal Objective, Proposal, Customer Response, and Next Action." The summarization processing unit 212 will include the text of the deal information in the prompt.

[0032] The summarization processing unit 212 inputs the example data to the large-scale language model, including it in the prompt, if the summarization condition includes example data for fusion learning. The example data includes an example of opportunity information and an example of a summary result corresponding to that opportunity information. The summarization processing unit 212 includes, for example, an explanatory statement such as "The following is an example of opportunity information and its summary" in the prompt, followed by the example of opportunity information and the example of a summary result included in the example data. After the example data, the summarization processing unit 212 includes an instruction such as "Now, summarize the following opportunity information in the same format" and the opportunity information to be summarized in the prompt.

[0033] The summarization processing unit 212 inputs the generated prompt to the first large-scale language model stored in the model storage unit 235, causing it to generate a summary result. The summarization processing unit 212 retrieves the summary result output from the large-scale language model. The summarization processing unit 212 stores the retrieved summary result in the deal history storage unit 234. The summarization processing unit 212 can send the summary result to the output unit 219 for display on the user terminal 1.

[0034] The summarization processing unit 212 can also include a prompt that instructs the user to extract the summary item selected by the user from the sales opportunity information, presents the detected summary item candidates to the user, and extracts the summary item selected by the user from the sales opportunity information. For example, the summarization processing unit 212 analyzes the text of the sales opportunity information and detects keywords such as "purpose," "proposal," "response," and "action." The summarization processing unit 212 displays the detected keyword candidate summary item on the user terminal 1 and prompts the user to make a selection. The summarization processing unit 212 includes the summary item selected by the user in the prompt.

[0035] The summarization processing unit 212 can also include a prompt that instructs the first large-scale language model to include, in addition to the summarization results, candidates for next actions and candidates for risks extracted based on the opportunity information. For example, the summarization processing unit 212 might include a prompt such as, "In addition to the summary, please extract candidates for next actions and candidates for risks." This allows the system to obtain information about the next steps and risks of the opportunity along with the summarization results. Note that the information extracted or generated by the summarization processing unit 212 in addition to the summarization results is not limited to candidates for next actions and candidates for risks, but may also include the results of an information extraction task that extracts the job title of the opportunity partner, etc., or the results of a text generation task that generates the outline of the opportunity, etc. (See Modification 11).

[0036] The summarization processing unit 212 can also input a second prompt, which includes the summarization result and opportunity information generated by the first large-scale language model, to the second large-scale language model to generate a justification or supplementary explanation for the summarization result. For example, the summarization processing unit 212 generates a second prompt, which includes the summary result and opportunity information, along with an instruction such as, "Please explain which part of the opportunity information the following summary result was based on." The summarization processing unit 212 inputs the second prompt to the second large-scale language model to obtain a justification or supplementary explanation. This can improve the reliability of the summary result. Note that the large-scale language models used by the summarization processing unit 212 are not limited to the first and second large-scale language models, but may use three or more large-scale language models. Also, the content generated by the second large-scale language model is not limited to justification or supplementary explanation. For example, the summarization processing unit 212 may instruct the second large-scale language model to perform a process to suppress expressions in the summary result. Specifically, the summarization processing unit 212 may cause the second large-scale language model to determine whether the summarization result generated by the first large-scale language model contains useful information, and for items that are determined not to contain useful information, it may overwrite the summarization result to indicate that there was no mention of that item.

[0037] The summarization processing unit 212 can also include participant information and date information attached to the deal information in the prompt. For example, the summarization processing unit 212 might include information such as "Deal date and time: January 15, 2024, Participants: Mr. B from Company A, Mr. D from Company C" in the prompt. This allows the context of the deal to be conveyed to the large-scale language model, resulting in more appropriate summarization results.

[0038] The evaluation unit 213 evaluates the quality of the summary result generated by the summarization processing unit 212. The evaluation unit 213 evaluates the summary result based on evaluation indicators stored in the evaluation condition storage unit 233. For example, the evaluation unit 213 checks whether keywords or key phrases that should be included in the summary result are actually included. The evaluation unit 213 checks whether the length of the summary result conforms to the evaluation conditions. The evaluation unit 213 evaluates the comprehensiveness and conciseness of the summary result. In addition to, or instead of, the evaluation unit 213 may evaluate the quality of the summary result using a large-scale language model. In this case, the evaluation unit 213 inputs a prompt including the summary result and evaluation indicators to the large-scale language model, and obtains the evaluation result by having the large-scale language model evaluate the quality of the summary result.

[0039] The evaluation unit 213 can calculate the evaluation results as a score. For example, the evaluation unit 213 calculates a score from 0 to 100 for each evaluation item and calculates the average of the scores for each evaluation item as the overall score. The evaluation unit 213 stores the calculated score in the negotiation history storage unit 234. If the score falls below a predetermined threshold, the evaluation unit 213 determines that the quality of the summary result is low.

[0040] The summary condition update unit 218 updates the summary conditions corresponding to the company based on the evaluation results from the evaluation unit 213. The summary condition update unit 218 modifies at least a portion of the summary items of the summary conditions or the example data for fusion learning, depending on the degree of fit to the evaluation conditions. For example, if a particular keyword is not included in the summary result, the summary condition update unit 218 adds a summary item related to that keyword to the summary conditions. If the length of the summary result does not conform to the evaluation conditions, the summary condition update unit 218 modifies the summary conditions to add instructions regarding the length of the summary to the prompt.

[0041] The summary condition update unit 218 can also update the summary conditions corresponding to the company to which a user belongs in response to an operation input from a user whose authority type stored in the user information storage unit 232 meets predetermined conditions. For example, the summary condition update unit 218 can receive a request to change the summary conditions from a user with administrator privileges and update the summary conditions stored in the summary condition storage unit 231. The summary condition update unit 218 can be configured not to accept change requests from general users.

[0042] The progress prediction unit 214 predicts the future progress of business negotiations related to the company based on the summary results and negotiation information. The progress prediction unit 214 makes predictions using a machine learning model. The progress prediction unit 214 takes negotiation information, summary results, and past negotiation history information related to the company as input and outputs prediction results that include at least one of the probability of closing the deal and the expected closing date.

[0043] The progress prediction unit 214 retrieves past sales negotiation history information related to the company from the sales negotiation history storage unit 234. The progress prediction unit 214 retrieves past sales negotiation information, summary results, sales negotiation progress status, and closing results. The progress prediction unit 214 inputs this information as features into a machine learning model. The machine learning model is a model trained using past sales negotiation data and is trained to predict the probability of closing a deal and the expected closing time based on the characteristics of the sales negotiation.

[0044] The progress prediction unit 214 acquires the prediction results output from the machine learning model. The prediction results include, for example, a percentage indicating the probability of closing a deal and a date indicating the expected closing date. The progress prediction unit 214 stores the prediction results in the deal history storage unit 234. The progress prediction unit 214 can send the prediction results to the output unit 219 and display them on the user terminal 1.

[0045] The improvement plan generation unit 215 generates an improvement plan, including improvement measures related to business negotiations with the company, based on the prediction results and summary results from the progress prediction unit 214. The improvement plan generation unit 215 generates the improvement plan using a large-scale language model. The improvement plan generation unit 215 generates a prompt including the summary results, the prediction results from the progress prediction unit 214, and indicators calculated by the continuous monitoring unit 216, and inputs it into the large-scale language model.

[0046] The improvement plan generation unit 215 generates an improvement plan that includes at least part of the proposed agenda for the next business meeting, the person to follow up with, and points to review the proposal. The improvement plan generation unit 215 includes prompts such as, "Based on the following business meeting summary and prediction results, please create a proposed agenda for the next business meeting. Also, please suggest the person to follow up with and points to review the proposal." The improvement plan generation unit 215 retrieves the improvement plan output from the large-scale language model.

[0047] The improvement plan generation unit 215 stores the generated improvement plan in the sales negotiation history storage unit 234. The improvement plan generation unit 215 transmits the improvement plan to the output unit 219, which can then display it on the user terminal 1. The user can use the displayed improvement plan as a reference to prepare for the next sales negotiation.

[0048] The continuous monitoring unit 216 records summary results and progress forecast results from the progress forecast unit 214 for each company, and calculates indicators related to the business negotiations based on these records. The continuous monitoring unit 216 calculates business negotiation indicators for each company, including at least a portion of the number of business negotiations, closing rate, proposed amount, and number of risk items included in the summary results within a predetermined period.

[0049] The continuous monitoring unit 216 retrieves negotiation history information for each company from the negotiation history storage unit 234. The continuous monitoring unit 216 counts the number of negotiations within a predetermined period. The continuous monitoring unit 216 calculates the ratio of the number of concluded negotiations to the total number of negotiations and uses this as the closing rate. The continuous monitoring unit 216 calculates the total proposed amount. The continuous monitoring unit 216 counts the number of risk items included in the summary results. The continuous monitoring unit 216 stores the calculated negotiation indicators in the negotiation history storage unit 234.

[0050] The adaptive control unit 217 updates at least one of the summary conditions stored in the summary condition storage unit 231 or the generation logic of the improvement plan generation unit 215 based on the indicators and prediction results calculated by the continuous monitoring unit 216. If the sales opportunity indicator falls below a predetermined threshold, the adaptive control unit 217 automatically changes at least a portion of the summary items of the summary conditions corresponding to the company or the example data for Future Shot learning.

[0051] The adaptive control unit 217 adds summary items such as "approver information" and "budget size" to the summary conditions if, for example, the closing rate falls below a predetermined threshold. The adaptive control unit 217 also adds a summary item called "risk countermeasures" to the summary conditions if the number of risk items exceeds a predetermined threshold. The adaptive control unit 217 stores the modified summary conditions in the summary condition storage unit 231.

[0052] The adaptive control unit 217 can also add negotiation information and summary results to the training dataset of the machine learning model in the progress prediction unit 214 based on the error between the prediction result of the progress prediction unit 214 and the actual closing result, and retrain the machine learning model. If the error between the prediction result and the actual closing result exceeds a predetermined threshold, the adaptive control unit 217 determines that the accuracy of the machine learning model has decreased. The adaptive control unit 217 adds new negotiation information and summary results to the training dataset and retrains the machine learning model. This improves the prediction accuracy of the machine learning model.

[0053] The adaptive control unit 217 can acquire user evaluation information on the improvement plan and, based on this evaluation information, update at least a portion of the prompt text or example data for fusion learning that is input to the large-scale language model in the improvement plan generation unit 215. The adaptive control unit 217 receives evaluations of the improvement plan from the user, for example, via the user terminal 1. The evaluation can be, for example, a 5-point scale, a good / bad binary evaluation, or free-form comments. If the evaluation is low, the adaptive control unit 217 changes the prompt text or adds example data. This improves the quality of the improvement plan.

[0054] The output unit 219 transmits and displays summary results, forecast results, improvement plans, sales opportunity indicators, etc., to the user terminal 1. The output unit 219 transmits this information in response to requests from the user terminal 1. The output unit 219 can also generate a user interface for displaying the information and transmit it to the user terminal 1. The user interface can be, for example, a web page or an application screen.

[0055] Figure 4 illustrates the summarization processing flow of business negotiation information in an information processing system.

[0056] The business negotiation information acquisition unit 211 acquires audio data or text data related to the business negotiation (S101). If the data is audio, the business negotiation information acquisition unit 211 performs speech recognition processing and generates business negotiation information consisting of text data (S102). The summarization processing unit 212 identifies the company related to the business negotiation information (S103) and acquires the summarization conditions corresponding to that company from the summarization condition storage unit 231 (S104). The summarization processing unit 212 generates a prompt including the acquired summarization conditions and business negotiation information (S105), and inputs the prompt into a large-scale language model to generate a summary result (S106). The summarization processing unit 212 stores the generated summary result in the business negotiation history storage unit 234 (S107) and transmits it to the user terminal 1 via the output unit 219 (S108).

[0057] Figure 5 illustrates the update process flow for summary conditions in an information processing system.

[0058] The evaluation unit 213 acquires the summary result generated by the summarization processing unit 212 (S201) and evaluates the quality of the summary result based on the evaluation index stored in the evaluation condition storage unit 233 (S202). The evaluation unit 213 calculates the evaluation result as a score (S203) and determines whether the score falls below a predetermined threshold (S204). If the score falls below the threshold (YES in S204), the summary condition update unit 218 modifies the summary items or example data of the summary condition based on the evaluation result (S205) and stores the modified summary condition in the summary condition storage unit 231 (S206). If the score is equal to or greater than the threshold (NO in S204), the process ends.

[0059] Figure 6 illustrates the processing flow for predicting the progress of business negotiations and generating improvement plans in an information processing system.

[0060] The progress prediction unit 214 acquires the summary result and negotiation information (S301), and acquires past negotiation history information related to the company from the negotiation history storage unit 234 (S302). The progress prediction unit 214 inputs this information into a machine learning model and generates a prediction result including the probability of closing the deal and the expected closing time (S303). The continuous monitoring unit 216 calculates negotiation indicators for each company (S304). The improvement plan generation unit 215 generates a prompt including the summary result, prediction result, and negotiation indicators (S305), and inputs the prompt into a large-scale language model to generate an improvement plan (S306). The improvement plan generation unit 215 stores the generated improvement plan in the negotiation history storage unit 234 (S307) and transmits it to the user terminal 1 via the output unit 219 (S308).

[0061] Figure 7 illustrates the adaptive control processing flow in an information processing system.

[0062] The adaptive control unit 217 acquires the deal indicator calculated by the continuous monitoring unit 216 (S401) and determines whether the deal indicator falls below a predetermined threshold (S402). If the deal indicator falls below the threshold (YES in S402), the adaptive control unit 217 automatically changes the summary items or example data of the summary conditions (S403) and stores the changed summary conditions in the summary condition storage unit 231 (S404). The adaptive control unit 217 calculates the error between the prediction result of the progress prediction unit 214 and the actual closing result (S405) and determines whether the error exceeds a predetermined threshold (S406). If the error exceeds the threshold (YES in S406), the adaptive control unit 217 adds the deal information and summary result to the learning dataset and retrains the machine learning model (S407). If the deal indicator is above the threshold (NO in S402) or the error is below the threshold (NO in S406), the process ends.

[0063] As described above, the information processing system of this embodiment can generate summaries of business negotiation information based on summarization conditions set for each company, thereby enabling the efficient acquisition of appropriate business negotiation summaries tailored to each company's business characteristics and summarization needs. Furthermore, by using a large-scale language model, natural and easy-to-read summaries can be automatically generated. In addition, by using example data for fusion learning, summaries can be generated that conform to the format and content required by the company.

[0064] Furthermore, the information processing system of this embodiment can evaluate the quality of the generated summary results and automatically update the summary conditions based on the evaluation results, thereby continuously improving the quality of the summaries. This reduces the effort required for users to manually adjust the summary conditions. Note that the automatic updating of the summary conditions described above is not limited to being performed solely based on evaluations by the evaluation unit 213 based on evaluation indicators. The updating of the summary conditions may also be performed taking into account user feedback, as shown in Modification 4, or it may also be performed taking into account the results of monitoring or evaluation by another large-scale language model.

[0065] Furthermore, according to the information processing system of this embodiment, it is possible to predict the progress of business negotiations and generate improvement plans based on the prediction results, thereby improving the efficiency of sales activities and increasing the closing rate. The continuous monitoring unit calculates indicators across multiple business negotiations, and the adaptive control unit can automatically update the summarization conditions and prediction models, thereby continuously improving the accuracy of the entire system.

[0066] Furthermore, the information processing system of this embodiment can acquire business negotiation information in a variety of ways, such as acquiring negotiation information from voice data, linking with external negotiation management systems, and acquiring data from online meeting systems, thus enabling it to adapt to various work environments. This enhances user convenience.

[0067] Although these embodiments have been described above, they are intended to facilitate understanding of the present invention and are not intended to limit its interpretation. The present invention can be modified and improved without departing from its spirit, and equivalents thereof are also included.

[0068] For example, the processing performed by each functional unit of the management server 2 described above may be executed by any of the functional units. Furthermore, different functional units may be added to perform some of the processing performed by each of the functional units described above. Also, the functional units of the management server 2 may be distributed across multiple computers.

[0069] Furthermore, the information stored in each memory unit of the management server 2 may be stored in any of the memory units. That is, the information stored in the multiple memory units mentioned above may be stored in a single memory unit, or a portion of the information stored in one memory unit may be stored in another memory unit.

[0070] <Example 1> In the embodiment described above, an example was shown in which the summarization processing unit 212 inputs prompts containing summarization conditions and deal information to a large-scale language model to generate a summary result. However, the summarization processing unit 212 may divide the deal information into multiple segments, generate a summary individually for each segment, and then integrate the summaries of each segment to generate a final summary result. Specifically, the summarization processing unit 212 divides the text of the deal information into segments of a predetermined number of characters or paragraphs. For each segment, the summarization processing unit 212 inputs it to the large-scale language model along with prompts containing summarization conditions to generate a summary for each segment. The summarization processing unit 212 then combines the summaries of each generated segment and inputs the entire combined summary to the large-scale language model to generate a final summary result. This makes it possible to generate a summary even for lengthy deal information without exceeding the input token limit of the large-scale language model.

[0071] <Modification 2> In the embodiment described above, an example was shown in which the summary condition storage unit 231 stores summary conditions for each company. However, the summary condition storage unit 231 may store different summary conditions for each department or each user within a company. Specifically, in addition to company identification information, the summary condition storage unit 231 stores summary conditions in association with department identification information or user identification information. The summary processing unit 212 obtains the department information or user information of the user associated with the business opportunity information from the user information storage unit 232 and obtains the summary conditions corresponding to that department or user from the summary condition storage unit 231. The summary processing unit 212 generates a summary result using the obtained summary conditions. This makes it possible to set different summary items for sales departments and technical departments, or to generate summaries with different levels of detail for managers and general employees, even within the same company.

[0072] <Variation 3> In the embodiment described above, an example was shown in which the negotiation information acquisition unit 211 acquires negotiation information from audio data or text data. However, the negotiation information acquisition unit 211 may also acquire video data related to the negotiation, extract audio data and image data from the video data, and generate negotiation information by combining the text data generated from the audio data and the information extracted from the image data. Specifically, the negotiation information acquisition unit 211 acquires video data from an online meeting system or recording device. The negotiation information acquisition unit 211 extracts an audio track from the video data and performs speech recognition processing to generate text data. The negotiation information acquisition unit 211 extracts image frames from the video data and performs image recognition processing to extract information such as text, graphs, and charts from the displayed materials. The negotiation information acquisition unit 211 integrates the text data generated from the audio and the information extracted from the images in chronological order to generate negotiation information. This makes it possible to acquire comprehensive negotiation information, including visual information presented during the negotiation.

[0073] <Modification 4> In the embodiment described above, an example was shown in which the evaluation unit 213 evaluates the summary result based on evaluation indicators stored in the evaluation condition storage unit 233. However, the evaluation unit 213 may also acquire feedback from the user and use that feedback as an evaluation indicator. Specifically, the output unit 219 provides a user interface for inputting an evaluation of the summary result when displaying the summary result on the user terminal 1. The user can evaluate whether each item of the summary result is appropriate or not, or evaluate their overall satisfaction with the summary result. The evaluation unit 213 acquires evaluation information transmitted from the user terminal 1 and calculates a quality score of the summary result based on the evaluation information. The evaluation unit 213 can also aggregate evaluation information from multiple users and calculate an average score. The summary condition update unit 218 changes the summary conditions for summary items that have received low evaluations from users. This makes it possible to improve the summary conditions based on actual user satisfaction.

[0074] <Modification 5> In the embodiment described above, an example was shown in which the progress prediction unit 214 predicts the progress of a business negotiation using a machine learning model. However, the progress prediction unit 214 may also predict the progress of a business negotiation using a large-scale language model. Specifically, the progress prediction unit 214 generates a prompt that includes business negotiation information, a summary result, and past business negotiation history information. The progress prediction unit 214 includes an instruction in the prompt such as, "Predict the probability of closing this business negotiation and the expected closing date." The progress prediction unit 214 inputs the generated prompt into the large-scale language model and obtains the prediction result. The large-scale language model has learned patterns from past business negotiation history information and can infer the possibility of closing from the content of the business negotiation information. The progress prediction unit 214 parses the prediction result output from the large-scale language model and extracts the probability of closing and the expected closing date. This makes it possible to predict the progress of a business negotiation by utilizing the general inference capabilities of the large-scale language model without separately training a machine learning model. Furthermore, the progress prediction unit 214 may express the progress status of a business negotiation (e.g., probability of closing the deal) using a score obtained by adding or aggregating multiple indicator values ​​output by a large-scale language model based on the negotiation information. In other words, the progress prediction by the progress prediction unit 214 is not limited to using a machine learning model, but may also use the output results of a large-scale language model.

[0075] <Variation 6> In the embodiment described above, an example was shown in which the improvement plan generation unit 215 generates an improvement plan that includes a draft agenda for the next business negotiation, the person to follow up with, and points for reviewing the proposed content. However, the improvement plan generation unit 215 may also acquire industry information, competitor information, and market trend information of the business partner's company from an external database, and input prompts containing this information into a large-scale language model to generate an improvement plan. Specifically, the improvement plan generation unit 215 acquires industry information from an external company information database based on the identification information of the business partner's company. The improvement plan generation unit 215 acquires competitor information from a competitor analysis database. The improvement plan generation unit 215 acquires market trend information from a market research database. The improvement plan generation unit 215 generates prompts containing this information, as well as summary results and forecast results, and inputs them into a large-scale language model. The large-scale language model can generate a more strategic improvement plan that takes into account industry trends and the competitive situation. This makes it possible to propose specific and practical improvement measures to increase the probability of success in business negotiations.

[0076] <Example 7> In the embodiment described above, an example was shown in which the continuous monitoring unit 216 calculates negotiation indicators for each company. However, the continuous monitoring unit 216 may also calculate negotiation indicators for each type of negotiation or each product category. Specifically, the negotiation history storage unit 234 stores the type of negotiation (new negotiation, additional proposal to an existing customer, contract renewal, etc.) or product category (product A, service B, etc.) for each negotiation. The continuous monitoring unit 216 calculates indicators such as the number of negotiations, closing rate, and proposed amount for each type of negotiation or product category. The continuous monitoring unit 216 stores the calculated indicators in the negotiation history storage unit 234. The adaptive control unit 217 adjusts the summarization conditions based on the indicators for each type of negotiation or product category. For example, if the closing rate is low for a particular product category, the summarization conditions are changed to include more detailed technical information and pricing information in the summarization items for negotiations in that product category. This makes it possible to optimize the summarization conditions in a fine-grained manner according to the characteristics of the negotiation.

[0077] <Differentiation Example 8> In the embodiment described above, an example was shown in which the adaptive control unit 217 updates the generation logic of the summarization conditions or improvement plan generation unit 215. However, the adaptive control unit 217 may also fine-tune the large-scale language model itself. Specifically, the adaptive control unit 217 collects business opportunity information, summarization results, and user evaluation information accumulated for each company as a training dataset. The adaptive control unit 217 uses the collected training dataset to perform fine-tuning on the large-scale language model. Fine-tuning is a method of further learning the parameters of a pre-trained large-scale language model to adapt them to a specific task or domain. The adaptive control unit 217 stores the fine-tuned large-scale language model in the model storage unit 235. The summarization processing unit 212 generates summarization results using the fine-tuned large-scale language model. This makes it possible to generate higher-quality summaries adapted to the characteristics and terminology of business opportunities specific to each company.

[0078] <Modification 9> In the embodiment described above, an example was shown in which the summarization processing unit 212 generates a summary result. However, the summarization processing unit 212 may generate multiple summary results using multiple different large-scale language models, compare and evaluate each summary result, and select the optimal summary result. Specifically, the model storage unit 235 stores multiple different large-scale language models. The summarization processing unit 212 inputs prompts containing the same summarization conditions and opportunity information to each large-scale language model and obtains a summary result from each model. The evaluation unit 213 calculates a quality score for each summary result based on evaluation indicators stored in the evaluation condition storage unit 233. The summarization processing unit 212 selects the summary result with the highest quality score as the final summary result. The summarization processing unit 212 stores the selected summary result in the opportunity history storage unit 234 and transmits it to the user terminal 1 via the output unit 219. This makes it possible to leverage the strengths of multiple models and provide higher quality summary results.

[0079] <Variation 10> In the embodiment described above, an example was shown in which the deal information acquisition unit 211 acquires deal information. However, the deal information acquisition unit 211 may also acquire related information such as emails, chat messages, and shared documents sent and received before and after the deal, and integrate this related information into the deal information. Specifically, the deal information acquisition unit 211 acquires emails or chat messages related to the deal via the email server or chat system API. The deal information acquisition unit 211 identifies the relevant emails or chat messages based on the date and time of the deal and participant information. The deal information acquisition unit 211 acquires documents shared during the deal from the document management system. The deal information acquisition unit 211 integrates the audio or text data of the deal with the content of the acquired emails, chat messages, and shared documents to generate comprehensive deal information. The summarization processing unit 212 generates a summary result using the integrated deal information. This makes it possible to generate a comprehensive summary that includes not only the deal setting but also the context before and after the deal.

[0080] <Variation 11> In the embodiment described above, an example was shown in which the summarization processing unit 212 generates a summary result of deal information. However, the processing performed by the summarization processing unit 212 using the large-scale language model is not limited to the generation of summaries. For example, the summarization processing unit 212 may perform an information extraction task to extract predetermined information from deal information. Examples of information extraction tasks include extracting the job title of the deal partner, whether or not they have the authority to make decisions, the budget size, and the names of competing products from deal information. The summarization processing unit 212 may also perform a text generation task to generate new text based on deal information. Examples of text generation tasks include generating an outline for the next deal, a draft of a proposal, and a draft of a follow-up email. In other words, the summarization processing unit 212 (processing unit) may be configured to input conditions set for each company and deal information into the large-scale language model to generate processing results for at least one of summarization, information extraction, and text generation. In this case, the summarization condition storage unit 231 stores conditions corresponding to each task of summarization, information extraction, or text generation for each company, and the summarization processing unit 212 generates processing results using the conditions corresponding to the target task. This allows for the implementation of a variety of processes, not limited to summarization, within a unified framework, tailored to the specific needs of each company. Furthermore, the generation of improvement plans by the aforementioned improvement plan generation unit 215 may also be implemented by the summarization processing unit 212 (processing unit) as one aspect of the text generation task.

[0081] <Variation 12> In the embodiment described above, an example was shown in which the summarization processing unit 212 inputs the deal information to be summarized and the summarization conditions to a large-scale language model to generate a summary result. However, the summarization processing unit 212 may also generate a new summary result by referring to past summary results stored in the deal history storage unit 234. Specifically, the summarization processing unit 212 obtains summary results related to past deals associated with the deal to be summarized from the deal history storage unit 234 and inputs these past summary results into the large-scale language model, including them in the prompt. This makes it possible to generate a summary result that takes into account the context across multiple deals related to the same company or the same deal. Furthermore, the summarization processing unit 212 may store in the summary condition storage unit 231, as an example of a good summary, any summary result or part thereof that has received a high evaluation from the evaluation unit 213 or feedback from the user among the previously generated summary results, and use such examples as exemplary data for future summarizations. This makes it possible to automatically accumulate the characteristics of past good summaries and improve the quality of subsequent summaries. Furthermore, the summarization processing unit 212 is not limited to cases where the text data necessary for summarization is explicitly specified in advance. It may also dynamically acquire text data related to the business negotiation to be summarized by searching the business negotiation history storage unit 234 or other data sources, and use the acquired text data for summarization. This makes it possible to generate summarization results while dynamically supplementing information useful for summarization.

[0082] <Example 13> In the embodiment described above, an example was shown in which the summary condition storage unit 231 stores summary conditions for each company (our own company) to which the user requesting the creation of the summary belongs. However, the summary condition storage unit 231 may also store summary conditions for each company of the business partner. Specifically, the summary condition storage unit 231 stores the identification information of the business partner company in association with the summary conditions. The summary processing unit 212 identifies the business partner company based on participant information or company identification information attached to the business opportunity information, and obtains the summary conditions corresponding to the identified business partner company from the summary condition storage unit 231. The summary processing unit 212 generates a summary result using the obtained summary conditions. This makes it possible to set summary items to focus on for each business partner company (for example, items related to the business partner company's decision-making process and key persons) according to the industry to which the business partner company belongs and the business policy with that company.

[0083] <Example 14> In the embodiment described above, an example was shown in which the summary condition storage unit 231 stores summary conditions for each company. However, the summary condition storage unit 231 may store different summary conditions for each combination of the company to which the user requesting the creation of the summary belongs (the company) and the company of the other party in the business negotiation. Specifically, the summary condition storage unit 231 stores summary conditions associated with combinations of the company's identification information and the other party's identification information. The summary processing unit 212 identifies the company to which the user associated with the business negotiation information to be summarized belongs from the user information storage unit 232 to identify the company, and identifies the other party's company based on participant information or company identification information attached to the business negotiation information. The summary processing unit 212 then obtains the summary conditions corresponding to the identified combination of the company and the other party's company from the summary condition storage unit 231. The summary processing unit 212 generates the summary result using the obtained summary conditions. This allows different summary conditions to be used depending on the counterparty company, even within the same company. For example, it is possible to set summary items that were important in past transactions with a specific counterparty company only for business negotiations with that company, enabling the setting of detailed summary conditions that correspond to the relationship between the company and the counterparty company. The summary condition storage unit 231 may also use a summary condition corresponding to the company or a summary condition corresponding to the counterparty company if no summary condition corresponding to the combination of the company and the counterparty company is stored.

[0084] <Disclosure Items> Furthermore, this disclosure also includes the following configurations. [Item 1] A business negotiation information acquisition unit acquires business negotiation information that shows the details of the business negotiation, For each company, there is a summary condition storage unit that stores the summary conditions corresponding to that company, A summarization processing unit inputs the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, An information processing system equipped with the following features. [Item 2] The information processing system described in item 1, The aforementioned business negotiation information acquisition unit is an information processing system that acquires audio data related to business negotiations and obtains the aforementioned business negotiation information, consisting of text data, by performing speech recognition processing on the audio data. [Item 3] The information processing system described in item 1, The aforementioned business negotiation information acquisition unit is an information processing system that receives text data related to business negotiations as input and acquires said text data as said business negotiation information. [Item 4] The information processing system described in item 1, The system further includes a user information storage unit that stores company information identifying the company to which each user belongs. The summarization processing unit identifies the affiliated company information corresponding to a user associated with the business opportunity information to be summarized from the user information storage unit, obtains the summarization conditions corresponding to the identified affiliated company information from the summarization condition storage unit, and includes the obtained summarization conditions in the prompt, thereby forming an information processing system. [Item 5] The information processing system described in item 1, The aforementioned summary conditions include one or more summary items to be extracted from the deal information, The summarization processing unit includes in the prompt a description instructing the large-scale language model to summarize, including the summarization result for each of the summarization items, as part of the information processing system. [Item 6] The information processing system described in item 1, The summarization conditions include at least one example relating to the summarization result as illustrative data for fusion shot learning. The summarization processing unit is an information processing system that inputs the example data into the prompt and then inputs it into the large-scale language model. [Item 7] The information processing system described in item 1, The summary condition storage unit is an information processing system having a summary condition setting function that receives and stores summary conditions input by a user. [Item 8] The information processing system described in item 5, The summarization processing unit detects expressions corresponding to the summary items contained in the business opportunity information, presents the detected summary item candidates to the user, and includes a description in the prompt that instructs the system to extract the summary item selected by the user from the business opportunity information. [Item 9] On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, A step of inputting the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, An information processing method that enables execution of [the specified action]. [Item 10] On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, A step of inputting the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, A program to execute. [Item 11] A business negotiation information acquisition unit acquires business negotiation information that shows the details of the business negotiation, For each company, there is a summary condition storage unit that stores the summary conditions corresponding to that company, A summarization processing unit inputs the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, An evaluation condition storage unit that stores evaluation indicators for the summary results, A summary condition update unit that evaluates the quality of the summary result based on the evaluation conditions and updates the summary condition corresponding to the company based on the evaluation result, An information processing system equipped with the following features. [Item 12] The information processing system described in item 11, The evaluation condition storage unit stores evaluation conditions that include at least one of the keywords or key phrases that should be included in the summary result and conditions relating to the length of the summary result. The summary condition update unit is an information processing system that modifies at least a portion of the summary items of the summary condition or the example data for fusion learning according to the degree of fit to the evaluation condition. [Item 13] The information processing system described in item 11, The system further includes a user information storage unit that stores user information, including company affiliation information that identifies the company to which the user belongs, and the user's permission type, for each user. The summary condition update unit is an information processing system that updates the summary condition corresponding to the company to which a user belongs, in response to an operation input from a user whose authority type stored in the user information storage unit meets predetermined conditions. [Item 14] The information processing system described in item 11, The aforementioned business negotiation information acquisition unit cooperates with an external business negotiation management system to acquire text data, participant information, and date information related to business negotiations recorded in the business negotiation management system, uses the acquired text data as the business negotiation information, and includes the acquired participant information and date information in the prompt, thereby forming an information processing system. [Item 15] The information processing system described in item 11, The aforementioned business negotiation information acquisition unit acquires audio data and screen sharing data obtained from the online meeting system. The summarization processing unit generates business opportunity information by combining text data generated from the audio data and title information of documents included in the screen sharing data, and inputs the business opportunity information into the first large-scale language model along with the first prompt, in this information processing system. [Item 16] The information processing system described in item 11, The summarization processing unit inputs a second prompt, which includes the summarization result generated by the first large-scale language model and the business opportunity information, into the second large-scale language model, causing it to generate an explanation of the basis for or supplementary explanation of the summarization result. [Item 17] The information processing system described in item 11, Information processing system wherein the summarization processing unit includes in the first prompt a description instructing the first large-scale language model to include, in addition to the summarization result, candidates for the next action and candidates for risk items extracted based on the business opportunity information. [Item 18] On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, A step of inputting the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, A step of storing the evaluation indicators for the summary results, The steps include: evaluating the quality of the summary result based on the evaluation conditions, and updating the summary conditions corresponding to the company based on the evaluation results; An information processing method that enables execution of [the specified action]. [Item 19] On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, A step of inputting the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, A step of storing the evaluation indicators for the summary results, The steps include: evaluating the quality of the summary result based on the evaluation conditions, and updating the summary conditions corresponding to the company based on the evaluation results; A program to execute. [Item 20] A business negotiation information acquisition unit acquires business negotiation information that shows the details of the business negotiation, For each company, there is a summary condition storage unit that stores the summary conditions corresponding to that company, A summarization processing unit inputs the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, Based on the summary results and the business negotiation information, a progress forecasting unit predicts the progress of future business negotiations related to the company, An improvement plan generation unit generates an improvement plan, including improvement measures related to business negotiations with the company, based on the prediction results from the progress prediction unit and the summary results. For each of the aforementioned companies, a continuous monitoring unit records the summary results and the prediction results of the progress prediction unit for multiple business negotiations, and calculates indicators related to the business negotiations based on said records. An adaptive control unit updates at least one of the summary conditions stored in the summary condition storage unit or the generation logic of the improvement plan generation unit based on the indicators calculated by the continuous monitoring unit and the prediction results, An information processing system equipped with the following features. [Item 21] The information processing system described in item 20, The aforementioned progress prediction unit is an information processing system that uses a machine learning model to take the business negotiation information, the summary results, and past business negotiation history information related to the company as inputs and outputs a prediction result that includes at least one of the probability of closing the deal and the expected closing date. [Item 22] The information processing system described in item 20, The improvement plan generation unit is an information processing system that inputs prompts to the large-scale language model, including the summarization results, the prediction results of the progress prediction unit, and indicators calculated by the continuous monitoring unit, to generate an improvement plan that includes at least a portion of the agenda for the next business negotiation, the person to follow up with, and points for reviewing the proposed content. [Item 23] The information processing system described in item 20, The continuous monitoring unit calculates, for each company, negotiation indicators that include at least a portion of the number of negotiations, closing rate, proposed amount, and the number of risk items included in the summary results within a predetermined period. The adaptive control unit is an information processing system that, when the sales opportunity indicator falls below a predetermined threshold, automatically modifies at least a portion of the summary items of the summary conditions corresponding to the company or the example data for fusion shot learning. [Item 24] The information processing system described in item 20, The adaptive control unit is an information processing system that, based on the error between the prediction result of the progress prediction unit and the actual closing result, adds the negotiation information and the summary result to the training dataset of the machine learning model in the progress prediction unit and retrains the machine learning model. [Item 25] The information processing system described in item 20, The adaptive control unit acquires user evaluation information regarding the improvement plan, and updates at least a portion of the prompt text or example data for fusion learning that is input to the large-scale language model in the improvement plan generation unit based on the evaluation information. [Item 26] On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, A step of inputting the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, Based on the summary results and the business negotiation information, the steps include predicting the progress of future business negotiations related to the company, The steps include generating an improvement plan, including improvement measures related to business negotiations with the company, based on the prediction results from the progress prediction unit and the summary results, For each of the aforementioned companies, the steps include recording the summary results and the prediction results of the progress forecasting unit for multiple business negotiations, and calculating indicators related to the business negotiations based on the records, The steps include updating at least one of the summary conditions stored in the summary condition storage unit or the generation logic of the improvement plan generation unit based on the indicators calculated by the continuous monitoring unit and the prediction results, An information processing method that enables execution of [the specified action]. [Item 27] On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, A step of inputting the summarization conditions and the business opportunity information corresponding to the company related to the business opportunity information to be summarized into a large-scale language model to generate a summary result of the business opportunity information, Based on the summary results and the business negotiation information, the steps include predicting the progress of future business negotiations related to the company, The steps include generating an improvement plan, including improvement measures related to business negotiations with the company, based on the prediction results from the progress prediction unit and the summary results, For each of the aforementioned companies, the steps include recording the summary results and the prediction results of the progress forecasting unit for multiple business negotiations, and calculating indicators related to the business negotiations based on said records, The steps include updating at least one of the summary conditions stored in the summary condition storage unit or the generation logic of the improvement plan generation unit based on the indicators calculated by the continuous monitoring unit and the prediction results, A program to execute. [Explanation of symbols]

[0085] 1 User terminal 2 Management Server

Claims

1. A business negotiation information acquisition unit acquires business negotiation information that shows the details of the business negotiation, For each company, there is a summary condition storage unit that stores the summary conditions corresponding to that company, A user information storage unit stores company information that identifies the company to which each user belongs, A summarization processing unit identifies the affiliated company information corresponding to the user associated with the deal information to be summarized from the user information storage unit, obtains the summarization conditions corresponding to the identified affiliated company information from the summarization condition storage unit, and inputs the obtained summarization conditions and the deal information into a large-scale language model to generate a summary result of the deal information. An information processing system equipped with the following features.

2. The information processing system according to claim 1, wherein the business negotiation information acquisition unit acquires audio data relating to a business negotiation and performs speech recognition processing on the audio data to acquire the business negotiation information consisting of text data.

3. The information processing system according to claim 1, wherein the business negotiation information acquisition unit receives text data relating to a business negotiation as input and acquires the text data as business negotiation information.

4. The aforementioned summary conditions include one or more summary items to be extracted from the deal information, The information processing system according to claim 1, wherein the summarization processing unit includes in the prompt a description that instructs the large-scale language model to summarize, including a summary result for each of the summarization items.

5. The summarization conditions include at least one example relating to the summarization result as illustrative data for fusion shot learning. The information processing system according to claim 1, wherein the summarization processing unit inputs the example data into the prompt and inputs it to the large-scale language model.

6. The information processing system according to claim 1, wherein the summary condition storage unit has a summary condition setting function that receives and stores summary conditions input by a user.

7. The information processing system according to claim 4, wherein the summarization processing unit detects expressions corresponding to the summary items included in the business negotiation information, presents the detected summary item candidates to the user, and includes a description in the prompt that instructs the user to extract the summary item selected by the user from the business negotiation information.

8. On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, For each user, there is a step of storing company information that identifies the company to which the user belongs, The steps include: identifying the affiliated company information corresponding to the user associated with the deal information to be summarized; inputting prompts including the summarization conditions and deal information corresponding to the identified affiliated company information into a large-scale language model to generate a summary result of the deal information; An information processing method that enables execution of [the specified action].

9. On the computer, Steps include obtaining negotiation information that shows the details of the negotiation, For each company, there is a step of memorizing the summary conditions corresponding to that company, For each user, there is a step of storing company information that identifies the company to which the user belongs, The steps include: identifying the affiliated company information corresponding to the user associated with the deal information to be summarized; inputting prompts including the summarization conditions and deal information corresponding to the identified affiliated company information into a large-scale language model to generate a summary result of the deal information; A program to execute.

Images