JP-7854753-B1 - Information processing system, information processing method, information processing program, and AI agent

JP7854753B1JP 7854753 B1JP7854753 B1JP 7854753B1JP-7854753-B1

Abstract

[Problem] To provide an information processing system, information processing method, information processing program, and AI agent that can perform processing appropriate to the user's prompt, even if the user's prompt contains incomplete information. [Solution] In the information processing system 10, the AI agent includes: prompt analysis means that analyzes prompts input by a user and stores user information in association with the analysis results of the prompts; input pattern learning means that learns or refers to multiple input patterns of prompts by a user based on the analysis results of multiple prompts and assigns identification information to the multiple input patterns; alternative prompt generation means that generates an alternative prompt by replacing the prompt with identification information when the prompt matches or is similar to any of the multiple input patterns; and processing execution means that executes processing corresponding to the alternative prompt. [Selection Diagram] Figure 1

Inventors

西郷孝一
池田雅人
諸冨大樹

Assignees

株式会社D4All

Dates

Publication Date: 20260507
Application Date: 20250729

Claims (7)

An information processing system configured to include an AI agent, The aforementioned AI agent, A prompt analysis means that analyzes prompts entered by a user, associates and stores user information about the user with the results of the prompt analysis, An input pattern learning means that learns or references multiple input patterns of the prompts by the user based on the analysis results of multiple prompts, and assigns identification information to the multiple input patterns, An alternative prompt generation means that generates an alternative prompt by modifying or supplementing the prompt if the prompt entered by the user or any person related to the user who can be identified by the user information matches or is similar to any of the plurality of input patterns to which the identification information has been attached, and the prompt contains content that does not match the intent of the user or the person related, or content that is ambiguous, The system includes processing execution means for executing processing corresponding to the alternative prompt, The aforementioned prompts include prompts related to management figures, The input pattern learning means learns or references the multiple input patterns by taking into account the analysis results of the multiple prompts, as well as information related to sales promotion measures and store layouts. An information processing system characterized by the following:
The information processing system according to claim 1, The alternative prompt generation means further generates the alternative prompt by replacing the specific information in the prompt with general information that is more general than the specific information. An information processing system characterized by the following:
An information processing method that is performed using an AI agent, The aforementioned AI agent, A prompt analysis step that analyzes prompts entered by a user, associates and stores user information about the user with the results of the prompt analysis, An input pattern learning step which learns multiple input patterns of the prompts by the user based on the analysis results of multiple prompts, and assigns identification information to the multiple input patterns, An alternative prompt generation step in which, if the prompt entered by the user or any person related to the user who can be identified by the user information matches or is similar to any of the plurality of input patterns to which the identification information has been attached, and the prompt contains content that does not match the intent of the user or the person related, or contains ambiguous content, the prompt is modified or supplemented to generate an alternative prompt; The process execution step includes, at least, executing a process that performs the process corresponding to the aforementioned alternative prompt, The aforementioned prompts include prompts related to management figures, The input pattern learning step learns or references the multiple input patterns by considering the analysis results of the multiple prompts, as well as information about sales promotion measures and store layouts. An information processing method characterized by the following:
In the information processing method described in claim 3, The alternative prompt generation step further generates the alternative prompt by replacing specific information in the prompt with general information that is more general than the specific information. An information processing method characterized by the following:
A program for an information processing system configured to include an AI agent, The aforementioned AI agent, which is a computer, A prompt analysis means that analyzes prompts entered by a user, associates and stores user information about the user with the results of the prompt analysis, An input pattern learning means that learns or references multiple input patterns of the prompts by the user based on the analysis results of multiple prompts, and assigns identification information to the multiple input patterns, An alternative prompt generation means that generates an alternative prompt by modifying or supplementing the prompt if the prompt entered by the user or any person related to the user who can be identified by the user information matches or is similar to any of the plurality of input patterns to which the identification information has been attached, and the prompt contains content that does not match the intent of the user or the person related, or content that is ambiguous, This is configured to function as a processing execution means that executes the processing corresponding to the aforementioned alternative prompt. The aforementioned prompts include prompts related to management figures, The input pattern learning means learns or references the multiple input patterns by taking into account the analysis results of the multiple prompts, as well as information related to sales promotion measures and store layouts. An information processing program characterized by the following:
In the information processing program described in claim 5, The alternative prompt generation means further generates the alternative prompt by replacing the specific information in the prompt with general information that is more general than the specific information. An information processing program characterized by the following:
An AI agent that acts as a substitute for a human, and has the ability to learn and make decisions on its own, Computers, A prompt analysis means that analyzes prompts entered by a user, associates and stores user information about the user with the results of the prompt analysis, An input pattern learning means that learns or references multiple input patterns of the prompts by the user based on the analysis results of multiple prompts, and assigns identification information to the multiple input patterns, An alternative prompt generation means that generates an alternative prompt by modifying or supplementing the prompt if the prompt entered by the user or any person related to the user who can be identified by the user information matches or is similar to any of the plurality of input patterns to which the identification information has been attached, and the prompt contains content that does not match the intent of the user or the person related, or content that is ambiguous, This AI agent functions as a processing execution means for executing processing corresponding to the aforementioned alternative prompt. The aforementioned prompts include prompts related to management figures, The input pattern learning means learns or references the multiple input patterns by taking into account the analysis results of the multiple prompts, as well as information related to sales promotion measures and store layouts. The correspondence between the analysis results of the multiple prompts and the multiple input patterns is learned by machine learning. An AI agent characterized by the following.

Description

This invention relates to an information processing system, an information processing method, and an information processing program using an AI agent, as well as to an AI agent. Conventionally, systems have been proposed that include generative AI to support the creation of internal company articles, with the generative AI automatically performing tasks such as structuring sentences and correcting typographical errors (see, for example, Patent Document 1). Japanese Patent Publication No. 2025-47972 This is a schematic diagram showing an overview of the information processing system 10 according to this embodiment 1.This is a schematic diagram showing an overview of the information processing system 10 according to this second embodiment.This is a system configuration diagram showing an example of the configuration of the information processing system 10 according to this second embodiment.This figure shows an example of a prompt PRM and an alternative prompt RPRM.This is a schematic diagram showing an overview of the information processing system 10 according to this third embodiment.This is a system configuration diagram showing an example of the configuration of the information processing system 10 according to this third embodiment.This figure shows an example of a prompt PRM and an alternative prompt RPRM. <Embodiment 1> The information processing system 10 according to Embodiment 1 of the present invention will be described below with reference to the drawings. <System Overview> First, an overview of the information processing system 10 according to this embodiment 1 will be described using Figure 1. Figure 1 is a schematic diagram showing an overview of the information processing system 10 according to this embodiment 1. The information processing system 10 is an information processing system configured with an AI agent 12, the AI agent 12 comprising: prompt analysis means 12a that analyzes prompt PRM input by a user and stores the user information relating to the user and the analysis results of the prompt PRM; input pattern learning means 12c that learns or references multiple input patterns of prompt PRM by a user based on the analysis results of multiple prompt PRMs and assigns identification information to the multiple input patterns; alternative prompt generation means 12c that generates an alternative prompt RPRM by replacing the prompt PRM with identification information when a prompt PRM input by the user or a person related to the user who can be identified by the user information matches or is similar to one of the multiple input patterns; and processing execution means 12d that executes processing corresponding to the alternative prompt RPRM. According to the information processing system 10, even if the user's prompt contains incomplete information, it can perform processing appropriate to the user's prompt, thereby increasing user satisfaction. Furthermore, by reducing the number of tokens in the prompt, processing speed can be increased. Here, "prompt" refers to instructions or questions given by the user to the system. Examples include AI prompts given to generating AI, and command prompts that provide instructions through command statements. "Token" refers to the meaningful string of characters contained within the prompt. Furthermore, "user information relating to a user" refers to information that can identify a user. This includes, for example, the user's personal information (e.g., the user's name, ID, job title, etc.) as well as the user's affiliation information (e.g., the company the user works for, the department or group the user belongs to, etc.). Additionally, "persons related to the user who can be identified through user information" include individuals with the same job title as the user, those working for the same company as the user, those belonging to the same department or group as the user, and the user's colleagues and superiors. Furthermore, "input patterns" include, for example, frequent use of specific information (e.g., specific tokens), frequent input of specific prompts (questions or tasks), input of ambiguous content as prompts (e.g., content containing pronouns, abbreviations, industry jargon, and ambiguous words), and input of content that does not match the user's intent as prompts (e.g., content containing inaccurate phrases or content that does not fit the context of the prompt). "Identifying information" refers to information that allows multiple input patterns to be identified, such as numbers, letters, and symbols. Furthermore, the prompt PRM may include prompts related to management figures, and the input pattern learning means 12b may learn or refer to multiple input patterns by considering not only the analysis results of multiple prompt PRMs but also information related to sales promotion measures and store layouts. Here, "management figures" refer to numerical values that indicate a company's management status and are used to evaluate a company'