Information processing system, information processing method, and program

Abstract

To provide an information processing system efficiently performing work that seriously considers a business idea.SOLUTION: An information processing system includes a chat step which inputs a user prompt about a business idea to a large-scale language model, acquires an answer from the large-scale language model, and starts chatting. The chat step inputs to the large-scale language model a first user prompt and a system prompt including logic for selecting one of a plurality of frameworks about business in accordance with expressions contained in the first user prompt when the first user prompt is inputted, acquires an answer according to the selection framework selected in the large-scale language model, displays it to a user terminal screen, acquires an answer according to the selection framework as an answer to the second user prompt and beyond, and displays it to the user terminal screen.SELECTED DRAWING: Figure 7

Description

[Technical Field]

[0001] The present invention relates to an information processing system, an information processing method, and a program that support the realization of business ideas. [Background technology]

[0002] Patent Document 1 and the like describe a business strategy evaluation system that uses a computer system to evaluate a business strategy. In Patent Document 1, the business strategy is evaluated using trained models that correspond to various frameworks used for business strategy analysis (3C analysis, 4P analysis, SWOT analysis, etc.). [Prior art documents] [Patent documents]

[0003] [Patent Document 1] International Publication No. 2022 / 264344 Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, in a corporate entity consisting of a group of economically and organizationally related companies or a single company, employees of the corporate entity are often required to propose business ideas. In such cases, there is a need to streamline the work required to generate and deepen business ideas. However, Patent Document 1 and other documents evaluate business strategies. Patent Document 1 does not take into consideration supporting the efficiency of work required to generate and deepen business ideas. [Means for solving the problem]

[0005] An information processing system for solving the above problem is an information processing system having a processor, wherein the processor constitutes a chat module that accepts input of a user prompt related to a business idea, inputs the user prompt into a large-scale language model, obtains an answer from the large-scale language model, and conducts a chat; further, when a first user prompt is input, the chat module inputs the first user prompt and a system prompt including logic for selecting one of a plurality of business frameworks in accordance with the wording included in the first user prompt into the large-scale language model, obtains a first answer according to the selection framework selected in the large-scale language model, and displays it on a user terminal screen; and obtains second and subsequent answers according to the selection framework as the answer to second and subsequent user prompts, and displays them on the user terminal screen.

[0006] An information processing method for solving the above problem is an information processing method executed by an information processing system having a processor, and includes: a receiving step of receiving an input of a user prompt related to a business idea; and a chat step of inputting the user prompt into a large-scale language model, obtaining an answer from the large-scale language model, and conducting a chat. When a first user prompt is input, the chat step inputs the first user prompt and a system prompt including logic for selecting one of a plurality of business frameworks according to wording included in the first user prompt into the large-scale language model, obtains a first answer according to the selection framework selected in the large-scale language model, and displays it on a user terminal screen. Furthermore, second and subsequent answers according to the selection framework are obtained as answers to second and subsequent user prompts, and displayed on the user terminal screen. Furthermore, the information processing method is installed and executed on a computer such as a server.

[0007] According to the above configuration, when a user inputs a first user prompt regarding a business idea, the chat can proceed according to one selected framework (e.g., a first framework suitable for a new business, or a second framework suitable for an existing business) selected from among multiple business frameworks that is suitable for the input business idea. This allows even users who are not familiar with business frameworks to efficiently proceed with the work of examining a business idea.

[0008] In the above information processing system, information processing method, and program, the chat module may be configured to cause the large-scale language model to perform the following processes: generating a search statement from the first user prompt; obtaining a first search result in a dictionary database; and obtaining the first answer according to the selection framework in response to the first search result.

[0009] According to the above configuration, the framework can be selected more appropriately in a large-scale language model by using the contents of the search results in the dictionary database and the first user prompt.

[0010] In the information processing system, the information processing method, and the program, the dictionary database may be an annual report database that accumulates annual report information on corporate activities. According to the above configuration, by using the annual report database of the corporate entity to which the user belongs as the dictionary database, it is possible to select a framework taking into consideration the past corporate activities of the corporate entity to which the user belongs.

[0011] The above information processing system, information processing method, and program may further include a search module, which is configured to generate search keywords from the chat in the chat module, obtain second search results in a related information database to be searched, input the second search results into the large-scale language model, obtain the second search result answers, and display them on the user terminal screen.

[0012] According to the above configuration, when a chat session is started to discuss a business idea, search keywords are extracted from the user prompt and related information is automatically searched for. This makes it easy to obtain information related to the business idea. Furthermore, users can enter further search prompts in chat format to further search for related information.

[0013] The above information processing system, information processing method, and program may further include a note module that executes a process of obtaining a summary of the chat in the chat module via the large-scale language model, and a note database that stores the summary, and the chat module may be configured to display on the user terminal screen an add note object for receiving an instruction to execute a process of storing the summary in the note database.

[0014] According to the above configuration, when a chat session is interrupted to discuss a business idea, the user can operate the note adding object to summarize the conversation history and save the summary in the note database. The user can easily resume the chat session by reading the summary from the note database.

[0015] In the above information processing system, information processing method, and program, the chat module may be configured to display, on the user terminal screen, a map generation object for receiving an instruction to execute a process for generating a visualization map that visualizes the chat in the chat module, and when the process for generating the visualization map is executed, the visualization map is obtained via the large-scale language model and displayed on the user terminal screen. With the above configuration, the contents of the business idea that have been considered so far can be represented using the visualization map.

[0016] The above information processing system, information processing method, and program may further include a proposal module that generates a proposal whose content corresponds to the summary, and a proposal database that stores the proposal, wherein the note module displays a proposal generation object on the user terminal screen to receive instructions to execute the process of generating the proposal from the summary, and the proposal module may be configured to set the status to either public, which means that the proposal can be viewed from the user terminal, or private, which means that the proposal cannot be viewed from the user terminal, and to store the status in the proposal database together with the proposal to be saved.

[0017] With the above configuration, once a business idea is finalized through chat, it can be saved as a summary in the note database, and then a proposal can be generated from this summary by operating the proposal generation object. If this proposal is set to be public, it can be viewed by other users and can be provided as reference material for other users.

[0018] The information processing system, the information processing method, and the program may further include a search module, the proposal database may manage evaluations of the proposals, and the search module may be configured to display on the user terminal screen proposal objects linked to the proposals in descending or descending order of the evaluations. With the above configuration, highly rated proposals can be easily found.

[0019] The above information processing system, information processing method, and program may further include a search module, which is configured to generate search keywords from the input prompt, obtain at least a third search result from the proposal database, input the third search result into the large-scale language model, obtain the answer, and display it on the user terminal screen.

[0020] According to the above configuration, search keywords can be generated from the input prompt, making the process of searching proposals more efficient. In addition, by inputting a new prompt based on the answer to the search result, a new answer can be obtained, and by repeating this process, desired search results can be more easily reached. [Effects of the Invention]

[0021] According to the present invention, even a user who is not familiar with business frameworks can efficiently create business ideas and conduct in-depth studies. [Brief explanation of the drawings]

[0022] [Figure 1] FIG. 1 is a diagram showing the configuration of an information processing system according to an embodiment. [Figure 2] FIG. 2 is a diagram showing the configuration of a user terminal that constitutes the information processing system according to the embodiment. [Figure 3] FIG. 3 is a diagram showing the configuration of the idea management server that constitutes the information processing system in the embodiment. [Figure 4] FIG. 4 is a diagram illustrating a database connected to the idea management server. [Figure 5] FIG. 5 is a diagram illustrating a database connected to the idea management server. [Figure 6] FIG. 6 is a functional block diagram of the idea management server. [Figure 7] FIG. 7 is a flowchart for chatting to a wall. [Figure 8] FIG. 8 is a diagram illustrating system prompts when chatting with a friend. [Figure 9] FIG. 9 is a diagram for explaining a search query generation function when searching the annual report database. [Figure 10] Figure 10 is a diagram explaining the Lean Canvas that is used when starting a new business. [Figure 11]Figure 11 is a diagram explaining the business model canvas used for existing businesses. [Figure 12] FIG. 12 is a diagram illustrating the selection of a framework. [Figure 13] FIG. 13(a) is a diagram showing a note list page when the note tab is selected on the top page, and FIG. 13(b) is a diagram showing a proposal list page when the proposal tab is selected. [Figure 14] FIG. 14(a) is a diagram showing a state in which the chat tab is selected on an ideation page for starting a chat to be used for chatting, and FIG. 14(b) is a diagram showing the state in the middle of a chat. [Figure 15] FIG. 15 shows the ideation page with the information search tab selected. [Figure 16] FIG. 16 is a diagram illustrating a prompt for outputting a mind map. [Figure 17] FIG. 17 is a diagram for explaining a mind map. [Figure 18] FIG. 18 is a diagram illustrating a note page. [Figure 19] FIG. 19 is a diagram illustrating the proposal creation page. [Figure 20] FIG. 20 is a diagram illustrating the proposal page. [Figure 21] FIG. 21(a) is a diagram illustrating a search list page when the search tab is selected, and FIG. 21(b) is a diagram illustrating a chat search page when the chat tag is selected. DETAILED DESCRIPTION OF THE INVENTION

[0023] An information processing system and an information processing method to which the present invention is applied will be described below with reference to the drawings. <Summary> An information processing system to which the present invention is applied is, for example, a system operated by a business entity consisting of a group of economically and organizationally related companies or a single company. A user who is an employee of the business entity uses the system to find business ideas related to the business entity's operations. The user uses the system to further consider and concretize the business idea they have conceived. The system uses a large-scale language model for bouncing ideas. Bouncing ideas is a process of sharing ideas, proposals, plans, etc. with others and seeking their opinions and feedback. The system uses chat for bouncing ideas. The user can chat with a virtual mentor who acts as a guide or advisor in the business to organize their thoughts. This allows the user to concretize their business idea. The chat used for bouncing ideas is conducted in accordance with a business framework appropriate for the user's business idea. The framework is not particularly limited, but an example is the Lean Canvas (first framework), which is appropriate for new businesses. Another example is the Business Model Canvas (second framework), which is appropriate for existing businesses.

[0024] While chatting, users can efficiently obtain relevant information related to the business idea they are considering by searching related information databases such as research databases and company databases. Users can then continue chatting while referring to the obtained related information. Chats can also be interrupted. When interrupted, a summary of the conversation history so far can be saved in a notebook. An interrupted chat can be resumed by reading the saved summary. With this system, as the discussion progresses or reaches a certain stage, a visualization map can be generated using the chat conversation history. Visualization maps are not particularly limited, but include, for example, mind maps.

[0025] When the chat finishes, the user can use the summary to generate a proposal. When generating a proposal, the user can set the proposal's status to either "public," which allows other users to view it, or "private," which prevents other users from viewing it. When the proposal's status is set to "public," the proposal will be included in searches performed using this system. In addition, third-party ratings are managed for each proposal. Therefore, multiple proposals stored in the database are displayed in a list in order of highest or lowest rating.

[0026] <Configuration of information processing system> As shown in FIG. 1, the information processing system 1 includes a user terminal 2, an idea management server 3, and an external server 4, which are interconnected via a network 5 so as to be able to communicate with each other. The user terminal 2 is a mobile terminal such as a tablet or smartphone compatible with a mobile communication system, which is managed by a user such as an employee of a company. Applications such as a web page browser and a program for idea management are installed on the user terminal 2. The idea management server 3 is a server managed by the company to which the user belongs, and is accessed from the user terminal 2. The idea management server 3 manages the user's ideas.

[0027] The external server 4 includes a large-scale language model 6 and a chat completion API (Chat Completions API) 7 as a chat completion module. The large-scale language model 6 learns from a large-scale text corpus and is used to perform natural language understanding tasks. The large-scale language model 6 has the function of interpreting sentences given in prompts and generating appropriate responses in that context. The chat completion API 7 has the function of, for example, processing input sentences to generate prompts, and inputting these prompts into the large-scale language model 6 to obtain answers. The chat completion API 7 complements chats between the user terminal 2 and the large-scale language model 6.

[0028] <User device> 2, the user terminal 2 is communicatively connected to the idea management server 3 via a network 5. The user terminal 2 is connected to the network 5 by communicating with communication devices such as a wireless base station compatible with communication standards such as 4G, 5G, and LTE (Long Term Evolution), and a wireless LAN (Local Area Network) router compatible with wireless LAN standards such as IEEE (Institute of Electrical and Electronics Engineers) 802.11.

[0029] The user terminal 2 includes a communication (interface) IF 11, a touch panel 12 serving as an input unit and an output unit, a memory 13, a storage unit 14, and a control unit 15. The communication IF 11 is an interface for inputting and outputting signals so that the user terminal 2 can communicate with external devices. The touch panel 12 is a device that combines a display panel, such as a liquid crystal display panel or an organic EL panel, with a touchpad. The touch panel 12 constitutes the user terminal screen. The touch panel 12 includes a display panel as a display unit that displays images and a touchpad as an input unit that accepts operations. The input unit may be a keyboard, a mouse, etc. The output unit may be a display panel without a touchpad. The input unit may also be configured to accept voice input using a microphone. The memory 13 temporarily stores programs, data processed by the programs, etc., and is a volatile memory such as a DRAM (Dynamic Random Access Memory). The storage unit 14 is storage for saving data, such as a flash memory or an HDD (Hard Disc Drive). The control unit 15 is hardware that executes an instruction set described in a program and is a processor composed of an arithmetic unit, registers, peripheral circuits, etc.

[0030] <Idea management server> 3, the idea management server 3 is a computer connected to a network 5. The idea management server 3 includes a communication IF 21, an input / output IF 22, a memory 23, a storage unit 24, and a control unit 25.

[0031] The communication IF 21 is an interface for inputting and outputting signals so that the idea management server 3 can communicate with external devices. The input / output IF 22 functions as an interface with an input device for receiving input operations from a user and an output device for presenting information to the user. The memory 23 is for temporarily storing programs, data processed by the programs, etc., and is a volatile memory such as a DRAM. The storage unit 24 is storage for saving data, such as a flash memory or HDD. The control unit 25 is hardware for executing an instruction set written in a program, and is a processor composed of an arithmetic unit, registers, peripheral circuits, etc.

[0032] <Database> 4 and 5, the idea management server 3 is connected to a plurality of databases. Specifically, the idea management server 3 includes a user database 31, a chat database 32, a note database 33, and a business plan database 34. The idea management server 3 also includes an annual report database 35, a research database 36, and a company database 37. All of the databases 31 to 37 may be included in the idea management server 3, or some of the databases may be provided on other external servers. For example, the user database 31, the chat database 32, the note database 33, and the business plan database 34 may be included in the idea management server 3, and the annual report database 35, the research database 36, and the company database 37 may be included on other external servers.

[0033] The user database 31 manages user information for users who have registered to use this system. For each user, the user database 31 manages a user ID that uniquely identifies the user, in association with the company name and department to which the user belongs, the user's name, email address, etc. Of course, the user database 31 may also manage other information, such as a setting to keep the user's name private, or a username, pen name, nickname, or the like that is different from the name to be made public instead of keeping the name private.

[0034] The chat database 32 manages user IDs, system prompts, previous summaries, questions (user prompts), answers, etc. in association with session IDs that uniquely identify chat sessions used for chatting. The previous summaries, questions (user prompts), answers, etc. constitute the conversation history.

[0035] The note database 33 manages user IDs, session IDs, text constituting the title of a summary, the summary, and other information in association with a note ID that uniquely identifies the note in which the summary is written. Here, a note is a saved summary of the chat history used for bouncing ideas, identified by a note ID. For example, when a chat is interrupted, the summary is saved in the note database 33, and when the chat is resumed, the summary is read out. The summary is updated each time the chat is interrupted and resumed. Furthermore, when the chat reaches a conclusion, the summary serves as the basis for generating a proposal.

[0036] The proposal database 34 manages, in association with a proposal ID that uniquely identifies a proposal, a proposal category ID that uniquely identifies whether the proposal content is a new business or an existing business, etc., a status that identifies whether the proposal is public or private, an evaluation of the proposal, the text that makes up the proposal title, a title image, an explanation, notes, a note ID, etc. When a proposal's status is set to public, it is included in the search targets of searches conducted by other users.

[0037] The annual report database 35 accumulates and manages the annual report information of businesses that use this system. Annual report information is distributed by businesses to shareholders, investors, financial institutions, etc. at the end of the fiscal year as part of information disclosure, and is a comprehensive report that includes information such as business strategies, financial status, and future vision. The information stored in the annual report database 35 can be used as a basis for determining whether a business idea that a user is considering is a new business or an existing business within the business. The annual report database 35 is a dictionary database for selecting frameworks suitable for new businesses and frameworks suitable for existing businesses.

[0038] The research database 36 is one or more databases selected from a technical paper database, a patent document database, a corporate research and development report database, a market research database, etc. This allows the user to search for technologies, market research, etc. related to the business idea being considered.

[0039] The company database 37 manages interview records (company names, business names, impressions, etc.) with emerging companies such as start-up companies. The company database 37 may also manage interview records (company names, business names, impressions, etc.) with existing companies such as major companies.

[0040] <Functional block diagram of idea management server> 6, the idea management server 3 includes a chat module 41, a note module 42, a proposal module 43, and a search module 44. These modules 41 to 44 are realized by the control unit 25 in the hardware configuration of the idea management server 3 executing a program.

[0041] The chat module 41 inputs user prompts related to business ideas into the large-scale language model 6 and acquires responses to the user prompts from the large-scale language model 6. Specifically, the chat module 41 executes processes for accepting user prompts and creating responses to the inputs when a user and the chat module 41 discuss a business idea. The chat module 41 stores system prompts. When a user inputs a business idea as a first user prompt, the system prompts include logic for selecting one of multiple business frameworks based on the wording included in the first user prompt. In this embodiment, the business frameworks include the Lean Canvas and the Business Model Canvas, and the large-scale language model 6 selects one of these frameworks. The definitions of each framework are also described. The chat module 41 also inputs a conversation history into the large-scale language model 6 and acquires a mind map based on the conversation history from the large-scale language model 6.

[0042] The note module 42 generates a summary from the conversation history of the chat (user prompts and replies) conducted by the chat module 41 via the large-scale language model 6, and saves the summary in the note database 33. When the chat is interrupted, the summary is saved in the note database 33, and the chat module 41 can resume the chat by reading out the summary. When consideration of a business idea has reached a certain stage, the summary saved in the note database 33 can be used as the basis for generating a proposal.

[0043] The proposal module 43 generates a proposal with content corresponding to the summary stored in the note database 33. The proposal module 43 sets a category for the content of the proposal to be generated, such as whether it is a new business or an existing business. The proposal module 43 sets the public / private status of the proposal. The proposal module 43 manages evaluations of the proposal. Furthermore, the proposal module 43 generates text, a title image, an explanation, notes, etc. that make up the title of the proposal.

[0044] When a chat session to bounce ideas off each other begins, the search module 44 extracts and generates search keywords from the user prompt, searches for information in related information databases such as the business proposal database 34, annual report database 35, research database 36, and company database 37, and obtains the text of the search results (second search results). The search results text is then input into the large-scale language model 6, and a formatted and summarized search result answer (second search results answer) is obtained. This makes it easy to obtain information related to the business idea. Furthermore, the user can enter further search prompts in chat format to further search for related information.

[0045] The search module 44 displays proposal objects linked to the proposals in descending or descending order of evaluation. Furthermore, the search module 44 searches for proposals separately from the chat used for bouncing ideas off the wall. Specifically, the search module 44 generates search keywords from the input prompt and answer, and acquires at least the text of the search results for proposals (third search results) in the proposal database 34. The search results for proposals are then input into the large-scale language model 6, and a formatted and summarized answer is acquired. This makes it possible to acquire related information while chatting for information search separately from the chat used for bouncing ideas off the wall.

[0046] <Chat processing procedure when hitting the wall> When a new chat session is started, the chat module 41 operates according to the flowchart shown in FIG.

[0047] When a user inputs a first user prompt regarding a business idea on the user terminal 2, the user terminal 2 displays the first user prompt on the user terminal screen and transmits the first user prompt to the idea management server 3. Then, in the idea management server 3, the chat module 41 issues a session ID. Then, the chat module 41 reads the first user prompt (step S1) and also reads a system prompt for selecting an appropriate business framework (step S2).

[0048] Figure 8 shows the system prompt. As frameworks, the Lean Canvas, which is suitable for new businesses, and the Business Model Canvas, which is suitable for existing businesses, are defined. The large-scale language model 6 is configured to select which framework is appropriate based on the first user prompt and the system prompt (see part A in Figure 8).

[0049] As shown in Figure 10, the Lean Canvas, which is used for new businesses, is a tool for concretizing a startup's business model. The Lean Canvas is structured to enable a startup to focus on important customers, issues, products, etc. (See part B in Figure 8).

[0050] As shown in Figure 11, the Business Model Canvas, which is used for existing businesses, is a tool for organizing and concretizing complex business structures. The Business Model Canvas is designed to help organize what value is proposed to what kind of customers, through what mediums and channels it will be provided, and how revenue will be generated (see part C in Figure 8).

[0051] When the chat module 41 sends the first user prompt to the large-scale language model 6, there is no conversation history up to that point, so the chat module 41 omits reading the conversation history from the chat database 32 (step S3). Then, the chat module 41 calls the chat completion API 7 (step S4) and sends the system prompt (see FIG. 8) and the first user prompt to the external server 4.

[0052] In the system prompt, in section D, the user is set to interact with a virtual mentor who is a professional in the Lean Canvas and Business Model Canvas. In section E, a user prompt is set as an idea the user is thinking of. In section F, the previous user prompts, answers, and summaries are set.

[0053] The chat completion API 7 executes the processes of steps S11 to S14 to select a framework. The chat completion API 7 executes the search statement generation function shown in FIG. 9 as necessary using a function_call function, which the chat completion API 7 automatically determines whether to execute (step S11). The search statement generation function is a function for generating a search statement for selecting a framework in response to the first user prompt. The chat completion API 7 generates a word search statement (search query) in the annual report database 35 from the wording included in the first user prompt. The chat completion API 7 then submits the search query to the annual report database 35 and obtains search results (first search results), for example, in text form (step S13). Note that the search here may include not only the annual report database 35 but also databases such as the company database 37 and the proposal database 34 as search target databases.

[0054] Next, the chat completion API 7 inputs the search result text and the system prompt into the large-scale language model 6, and causes the large-scale language model 6 to select either a framework based on the Lean Canvas or a framework based on the Business Model Canvas (step S13). For example, if the large-scale language model 6 finds a term related to the idea set in the search query in the annual report database 35, it is determined to be an existing business, and if not found in the annual report database 35, it is determined to be a new business. Then, the chat completion API 7 generates a first answer to the first user prompt based on the selected framework via the large-scale language model 6 (step S14). At the same time, the chat completion API 7 generates summaries of the first user prompt and the first answer via the large-scale language model 6.

[0055] The chat module 41 receives the first response from the chat completion API 7 and stores it in the chat database 32. Specifically, the chat module 41 stores the user ID, system prompt, first user prompt, first response, summary of the first user prompt and first response, etc. in association with the session ID as a conversation history in the chat database 32 (step S5). The chat module 41 also displays the first response on the user terminal screen (step S6).

[0056] As shown in Figure 12, at the time of the first response to the first user prompt, in the large-scale language model 6, based on the first user prompt and the search results in the annual report database 35, the Lean Canvas framework is selected for new businesses, and the Business Model Canvas framework is selected for existing businesses.

[0057] When the second user prompt is input at the user terminal in response to the first answer, the process is repeated from step S1. That is, the user terminal 2 displays the user prompts up to that point, the first answer, and the second user prompt on the user terminal screen to the extent that they can be displayed, and also transmits the second user prompt to the idea management server 3. Then, the chat module 41 reads the second user prompt (step S1). The chat module 41 reads the system prompt.

[0058] The chat module 41 then reads the summary of the first (previous) user prompt and the answers together with the second user prompt (step S3). Next, the chat module 41 calls the chat completion API 7 (step S4) and transmits the second user prompt, the system prompt, and the first summary to the external server 4.

[0059] Since either the Lean Canvas or the Business Model Canvas framework has already been selected, the chat completion API 7 skips step S12. Then, in accordance with the selected framework (step S13), the chat completion API 7 generates a second answer to the second user prompt (step S14) via the large-scale language model 6. At the same time, the chat completion API 7 generates a summary sentence up to the second answer via the large-scale language model 6.

[0060] The chat module 41 receives the second reply and the like from the chat completion API 7 and stores them in the chat database 32. Specifically, the chat module 41 stores the user ID, the second user prompt, the second reply, a summary of the responses up to the second reply, and the like in association with the session ID as a conversation history in the chat database 32 (step S5). The chat module 41 also displays the second reply on the user terminal screen (step S6). Thereafter, the chat module 41 repeats these processes.

[0061] As described above, when the first user prompt regarding a business idea is input, the chat can proceed according to one selected framework suitable for the input business idea, selected from multiple business frameworks (for example, a first framework (Lean Canvas) suitable for a new business, or a second framework (Business Model Canvas) suitable for an existing business). This allows even users who are not familiar with business frameworks to proceed with the work of examining a business idea according to an appropriate framework.

[0062] Furthermore, the selection of a framework in the large-scale language model 6 can be made more appropriate by using the contents of the search results in the annual report database 35 and the first user prompt. By using the annual report database 35 when selecting a framework, it is possible to select a framework taking into consideration the past business activities of the business entity to which the user belongs. For example, in the large-scale language model 6, if a term related to the idea set in the search query is found in the annual report database 35, a framework for an existing business is selected, and if not found in the annual report database 35, a framework for a new business is selected.

[0063] <User device screen transition> 13(a) and (b) show the top page 50 displayed on the user terminal screen of the user terminal 2. FIG.

[0064] FIG. 13(a) shows a note list page 50a in which the note tab 51a is selected on the top page 50, and FIG. 13(b) shows a proposal list page 50b in which the proposal tab 51b is selected on the top page 50. The bottom of the top page 50 is provided with an ideation object 56, a search object 57, and a settings object 58 as major classification objects. FIGS. 13(a) and 13(b) show the note list page 50a in which the ideation object 56 is selected. When the search object 57 is selected, the page transitions to a search page 100 as shown in FIG. 21. When the settings object 58 is selected, the page transitions to a settings page, details of which will be omitted.

[0065] 13(a), on the note list page 50a when the note tab 51a is selected, a chat start object 52 is displayed. When the chat start object 52 is operated, a new chat session is started. One or more note objects 53 are arranged below the chat start object 52.

[0066] The note object 53 is linked to one or more notes stored in the note database 33 and associated with the user ID of the user operating the user terminal 2. A summary identified by the note ID is stored in the note. When a specific note object 53 is selected, the summary stored as a note linked to the note object 53 is read from the note database 33. The notes linked to the listed note objects 53 are, for example, saved summaries of the conversation history when a user interrupts a chat. When a specific note object 53 is operated, the summary stored as that note is read, allowing the chat to be resumed.

[0067] As shown in FIG. 13(b), when the proposal tab 51b is selected, one or more proposal objects 54 are arranged on the proposal list page 50b. The proposal objects 54 are linked to one or more proposals associated with the user ID operating the user terminal 2, which are stored in the proposal database 34. When a specific proposal object 54 is selected, the proposal linked to the proposal object 54 is read from the proposal database 34. The proposals linked to the listed proposal objects 54 are, for example, proposals created by the user in the past, and are associated with the user ID.

[0068] FIG. 14(a) shows a chat page 60 for starting a chat used for hitting a message with someone, and FIG. 14(b) shows the chat in progress. The chat page 60 is the page to which the user transitions when the hitting a message start object 52 is operated on the top page 50 of FIG. 13(a). The chat page 60 has a hitting a message tab 61, an information search tab 62, and a note tab 63 at the top. When the hitting a message tab 61 is selected, a chat for hitting a message with someone can be conducted. The chat page 60 also has a session switching tab 64. The session switching tab 64 allows the user to select another chat session. The other selectable chat sessions are notes associated with the user ID.

[0069] Below the session switching tab 64 is a chat display object 64a that displays chats within the session. The bottom of the chat page 60 includes a prompt input object 65 for inputting a user prompt and a send object 66 for sending a user prompt. The bottom of the chat page 60 also includes a map generation object 67 for receiving an instruction to execute a process for generating a visualized map and a note addition object 68 for receiving an instruction to execute a process for adding a summary to a note. When the map generation object 67 is operated, the chat module 41 causes the large-scale language model 6 to generate a mind map based on the conversation history up to that point. When the note addition object 68 is operated, the note module 42 issues a note ID and saves a summary of the conversation history in the note database 33. When a chat read from the note database 33 is interrupted again, the note associated with the note ID is updated.

[0070] <Information search> FIG. 15 shows an information search page 69 with the information search tab 62 selected on the chat page 60.

[0071] Even when the information search tab 62 is selected, chatting as a chat can be continued by selecting the hit-to-the-wall tab 61. In the information search, if chat input has been made in advance on the hit-to-the-wall function, a search is simultaneously performed in the information search tab 62 using the same initial conditions (first user prompt). That is, the search module 44 extracts search keywords from the first user prompt of the hit-to-the-wall and automatically searches for them. Then, when the information search tab 62 is selected, the automatically searched information is displayed on the user terminal screen according to the information search prompt held by the search module 44.

[0072] The databases to be searched here are related information databases such as a proposal database 34, an annual report database 35, a research database 36, and a company database 37. Search keywords are generated by the search module 44 performing morphological analysis to extract nouns and verbs, and then converting the verbs into nouns to use as search keywords. Note that the search sentences may be generated by inputting a prompt entered by the user into the large-scale language model 6.

[0073] The search module 44 then inputs the search result text into the large-scale language model 6 to obtain a formatted and summarized search result answer. The search result answer is composed of a summary of the search result answer, a list, and a summary of supplementary explanations. Furthermore, the search result answer is provided with a link to the object of the search result item. When the object of the search result item is touched, the content stored in the related information database can be displayed on the user's terminal screen for confirmation.

[0074] In this way, when chatting about a business idea begins, search keywords are extracted from the first user prompt and related information is automatically searched for. Therefore, even in the early stages of the chat, information related to the business idea can be easily obtained by simply switching from the "Bubble-throwing" tab 61 to the "Information Search" tab 62. Furthermore, in the "Information Search" tab 62, further search prompts can be entered in chat format to further the search for related information.

[0075] <Visualization Map> Next, the operation when the map generating object 67 is operated will be described. FIG. 16 shows a prompt for outputting a mind map held by the chat module 41. When the map generation object 67 is operated, the chat module 41 reads the conversation history and the prompt for outputting a mind map and sends them to the external server 4. The chat completion API 7 then obtains a mind map based on the conversation history via the large-scale language model 6 (see FIG. 17). A mind map is a visualized map diagram and image that depicts free thinking, the flow of ideas, and information branching out from a central concept. The chat module 41 receives the image of the mind map from the chat completion API 7 and displays it on the user's terminal screen. This allows the contents of business idea considerations to be expressed using a visualized mind map.

[0076] <Notes> Next, the operation when the note addition object 68 (see FIG. 14) is operated will be described. 18 shows a note page 71. When the note addition object 68 is operated, the note module 42 reads the conversation history of the currently ongoing chat, a summary of the previous chat, etc. from the chat database 32, and transmits them to the external server 4. Then, the chat completion API 7 inputs a prompt for generating a summary of the conversation history into the large-scale language model 6, and acquires the summary. Alternatively, the note module 42 may read the conversation history of the currently ongoing chat, a summary of the previous chat, etc. from the chat database 32 and acquire them without going through the large-scale language model 6. Next, the note module 42 displays the note page 71 on the screen of the user terminal.

[0077] The note page 71 includes a title input object 72 and a note input object 73. Text constituting the title of the summary of the conversation history to be input to the note input object 73 is input to the title input object 72. The text constituting the title to be input to the title input object 72 is text data, and can be input via the input unit of the user terminal 2. The note input object 73 is input with the summary. Note that the text constituting the title may be generated from the summary input to the note input object 73 via the large-scale language model 6. Once the summary has been input to the note input object 73, the summary can also be modified as appropriate on the user terminal.

[0078] The note page 71 has a save object 74 and a proposal generation object 75 at the bottom. When the save object 74 is operated, the note module 42 issues a note ID and stores the user ID, session ID, text constituting the title of the summary, the summary, etc. in association with the note ID in the note database 33. When the note object 53 (see FIG. 13) is operated, the summary is read from the note database 33 and chat is resumed for use as a back-and-forth discussion. When the proposal generation object 75 is operated, the proposal module 43 generates a proposal with content corresponding to the summary stored in the note database 33.

[0079] In this way, when interrupting a chat session, the user can operate the note addition object 68 to summarize the conversation history and save the summary in the note database 33. The user can easily resume the chat session by reading the summary from the note database 33. The summary can also be used as the basis for a proposal.

[0080] <Proposal> Next, the operation when the proposal creation object 75 is operated will be described. 19 shows a proposal generation page 80. When the proposal generation object 75 is operated, the proposal module 43 displays the proposal generation page 80 on the user terminal screen. The proposal generation page 80 is based on the contents of the note page 71. The proposal generation page 80 includes a category selection object 81, a status selection object 82, a title input object 83, a description input object 84, a remarks input object 85, an attachment object 86, and a registration object 87.

[0081] The category selection object 81 allows the user to select whether the contents of the business proposal fall into a new business, an existing business, or other category. For example, when selecting a framework in a chat used for bouncing ideas off each other, if the Lean Canvas is selected for a new business, the new business may be automatically selected. When selecting a framework, if the Business Model Canvas is selected for an existing business, the existing business may be automatically selected. Note that the user can also select and modify the category via the user terminal 2 regardless of the result of the framework selection.

[0082] The status selection object 82 allows the user to set the proposal to either "public," which allows other users to view the proposal, or "private," which prevents other users from viewing the proposal. This selection is made through the user terminal 2. When the status is set to "public," the proposal is included in search results. The title input object 83 transcribes the text constituting the title entered in the title input object 72 of the note. Note that the text constituting the title can also be modified through the user terminal 2. The description input object 84 transcribes the summary entered in the note input object 73 of the note. This summary can also be modified through the user terminal 2. The remarks input object 85 allows the user to enter remarks, such as memos, related to the proposal through the user terminal 2. The attachment object 86 allows the user to upload desired image files, etc. The uploaded image files are saved in the proposal database 34. The image files are displayed as title images in the proposal object 54. The registration object 87 is operated when registering a proposal in the proposal database 34.

[0083] When the registration object 87 is operated, the proposal module 43 issues a proposal ID that uniquely identifies the proposal, and stores, in association with the proposal ID, a category ID, a status that identifies whether the proposal is public or private, a rating for the proposal, text constituting the title of the proposal, a title image, a description, notes, a note ID, and the like. A proposal registered in the proposal database 34 by the proposal module 43 is added to a list of proposal objects 54 on the proposal list page 50b when the proposal tab 51b is selected on the top page 50 (see FIG. 13(b)). This list of proposals is related to the user ID of the user. Third-party ratings are managed for each proposal. Therefore, the proposal objects 54 are displayed in a list in descending order of rating.

[0084] In this way, once a business idea has been finalized through chat, a proposal can be generated based on the notes stored in the note database 33. If this proposal is set to be public, it can be viewed by other users and can be provided as reference material for other users.

[0085] In FIG. 13(b), when one of the proposal objects 54 is operated from the list of proposals, a proposal page 90 shown in FIG. 20 is displayed on the user terminal screen. 20, the proposal page 90 includes a title display object 91, an explanation display object 92, and a related idea display object 93. The title display object 91 displays the text that constitutes the title entered in the title input object 83. The related idea display object 93 displays proposals that are highly related to the proposal in question.

[0086] The proposal module 43 vectorizes (embeds) the text constituting the title and the text of the description in the proposal, and then calculates the semantic relevance of each piece of information stored in the proposal database 34 (vectorized) using cosine similarity. The proposal module 43 then displays the text constituting the title of a proposal whose relevance exceeds a threshold as a related idea display object 93 with a link. As an example, the related idea display object 93 displays a predetermined number, such as the top three, of proposals whose relevance exceeds the threshold. Every time the related idea display object 93 is operated, the linked proposal can be read from the proposal database 34 and displayed on the user terminal screen.

[0087] <search> When the search object 57 is operated on the top page 50 (see FIGS. 13(a) and 13(b)), a search page 100 is displayed on the user terminal screen. The search page 100 includes a search tab 101 and a chat tag 102. FIG. 21(a) shows a search list page 103 in which the search tab 101 is selected, and FIG. 21(b) shows a chat search page 104 in which the chat tag 102 is selected. The search page 100 allows a search for proposals stored in the proposal database 34. The proposal objects 54 displayed here have a status set to "public." Furthermore, the search on the search page 100 differs from the search performed on the information search page 69 used for bouncing ideas off each other, and is a search for proposals that is unrelated to the chat used for bouncing ideas off each other. Of course, databases other than the proposal database 34 may also be included in the search targets.

[0088] When the search tab 101 is selected on the search list page 103, a list of proposal objects 54 is displayed. Each proposal object 54 includes an evaluation object 105. The evaluation object 105 is, for example, a heart mark. When the evaluation object 105 is touched once to light it up, the evaluation value is increased by 1. When the evaluation object 105 is touched again, the evaluation object 105 is turned off and the evaluation value is decreased by 1. The evaluation values ​​are managed in the proposal database 34. On the search list page 103, the proposals are initially listed in descending or ascending order of evaluation value. The search list page 103 also includes a search filter object 106. When the search filter object 106 is operated, a keyword for searching a proposal is input. After the keyword is input, the search module 44 executes a search and displays a list of proposal objects 54 containing the keyword. In this case, the proposal objects 54 may be displayed in descending or descending order of evaluation. Alternatively, the proposal objects 54 may be displayed in descending or descending order of relevance. After checking the contents of the proposal, the evaluation object 105 may be touched to rate it. According to the above configuration, the user can rate the proposals that he or she has checked, and can easily find proposals that have received high ratings.

[0089] Furthermore, the proposal module 43 may analyze the behavioral patterns of users who accessed the proposal and users who gave it a high rating, and group users with similar behavioral patterns. In this case, when one user in the group registers a proposal, a notification such as an e-mail informing the other users that the proposal has been registered may be automatically sent to their contact addresses, such as e-mail addresses, registered in the user database 31.

[0090] 21(b), a chat search page 104 allows users to search for proposals while chatting. The chat search page 104 includes a session selection tab 107, a prompt input object 108 for inputting a search prompt, and a transmission object 109 for transmitting a user prompt.

[0091] When a search prompt to start a search is entered, the search module 44 generates search keywords and searches for proposals in the proposal database 34. Here, the keywords are generated by extracting search keywords from the wording of the search prompt entered by the user. Specifically, morphological analysis is performed to extract nouns and verbs, and the verbs are converted to nouns to use as search keywords. The search keywords may also be generated by inputting the prompt entered by the user into the large-scale language model 6. The search module 44 then inputs the search results into the large-scale language model 6, obtains formatted and summarized search result answers, and displays them on the user terminal screen. The search result answers are composed of a summary of the search result answers, a list, and a summary of supplementary explanations. Furthermore, the search result answers are provided with links to the objects of the search result items. A list of proposal objects 54 is displayed on the user terminal screen.

[0092] According to the above configuration, search keywords can be generated from the input search prompt, making the process of searching proposals more efficient. In addition, by inputting a new prompt based on the answer to the search result, a new answer can be obtained, and by repeating this process, desired search results can be more easily reached.

[0093] <Effects of the embodiment> The above-described embodiment can provide the following effects. (1) The first user prompt regarding the business idea is entered. After that, the chat can proceed according to one selected framework that is suitable for the entered business idea, selected from multiple business frameworks (for example, the first framework (Lean Canvas) suitable for a new business, or the second framework (Business Model Canvas) suitable for an existing business). This allows even users who are not familiar with business frameworks to efficiently proceed with the work of examining business ideas.

[0094] (2) For example, in the large-scale language model 6, if a phrase related to the idea set in the search query is found in the annual report database 35, a framework for an existing business is selected, and if it is not found in the annual report database 35, a framework for a new business is selected. In this way, the framework selection in the large-scale language model 6 can be made more appropriate by using the content of the search results in the annual report database 35 and the first user prompt.

[0095] (3) When selecting a framework, by using the annual report database 35, it is possible to select a framework taking into consideration the past corporate activities of the corporate entity to which the user belongs.

[0096] (4) When chatting about business ideas begins, search keywords are extracted from the user prompts and related information is automatically searched for. This makes it easy to obtain information related to the business idea. Furthermore, users can enter further search prompts in chat format to further search for related information.

[0097] (5) When interrupting a chat session, the user can use the note addition object 68 to summarize the conversation history and save the summary in the note database 33. The user can easily resume the chat session by reading the summary from the note database 33. The summary can also be used as the basis for a proposal.

[0098] (6) The business ideas that have been considered so far can be represented using a mind map, which is a visualization map. (7) Once a business idea has been finalized through chat, a proposal can be generated based on the notes stored in the note database 33. If this proposal is set to be public, it can be viewed by other users and can be provided as reference material for other users.

[0099] (8) Users can rate the proposals they have reviewed and can easily find proposals that have received high ratings. (9) By generating search keywords from the input search prompt, the search process for proposals can be made more efficient. In addition, by inputting a new prompt based on the answers of the search results, a new answer can be obtained, and by repeating this process, the desired search results can be more easily reached.

[0100] The above-described embodiment can be modified as follows: The embodiment and the following modifications can be combined with each other within the scope of technical compatibility.

[0101] The information processing system 1 in this embodiment may omit the proposal module 43 that generates a proposal from notes. In this case, the proposal database 34 that manages proposals is also omitted. Furthermore, the search module 44 omits the proposal search function.

[0102] The information processing system 1 in this embodiment may omit the note module 42 that executes the process of acquiring a summary of the chat to be used for the chat message. In this case, the note database 33 that manages the notes may also be omitted.

[0103] In the information processing system 1 of this embodiment, the search module 44 may omit the function of searching the related database using conversations in chats used for hitting each other with a wall.

[0104] The framework may be a framework other than the Lean Canvas framework or the Business Model Canvas framework, or you may choose from three or more frameworks. Increasing the number of frameworks to choose from allows you to select a framework that better suits your business idea. In addition to the Lean Canvas and Business Model Canvas, frameworks such as SWOT analysis, Five Forces analysis, and IR frameworks may also be used.

[0105] The dictionary database used when selecting a framework may include databases such as the company database 37 and the business plan database 34 in addition to the annual report database 35. Also, a database other than the annual report database 35 may be used as the dictionary database. [Explanation of symbols]

[0106] 1. Information processing system 2...User terminal 3. Idea management server 4...External server 5. Network 6...Large-scale language models 7. Chat Complement API 50...Top page 50a...Note list page 50b…Proposal list page 51...Proposal tab 51a...note tab 51b…Proposal tab 52...Wall-hitting start object 53...Note object 54...Proposal object 60...Chat page 61...Hit the wall tab 62...Information Search Tab 63...Notes tab 67...Map generation object 68...Add Note Object 69...Information search page 71...note page 75…Proposal generation object 80…Proposal generation page

Claims

1. An information processing system including a processor, the processor configuring a chat module; The chat module: a module that accepts input of a user prompt regarding a business idea, inputs the user prompt into a large-scale language model, obtains an answer from the large-scale language model, and conducts a chat; Furthermore, the chat module When a first user prompt is input, the first user prompt and a system prompt including logic for selecting one of a plurality of business frameworks in accordance with a wording included in the first user prompt are input to the large-scale language model; Obtain a first answer according to the selection framework selected in the large-scale language model and display it on a user terminal screen; As the answer to the second or subsequent user prompt, a second or subsequent answer in accordance with the selection framework is obtained and displayed on the user terminal screen. Information processing system.

2. The chat module causes the large-scale language model to perform the following processes: generating a search statement from the first user prompt; obtaining a first search result in a dictionary database; and obtaining the first answer according to the selection framework in response to the first search result. The information processing system according to claim 1 .

3. The dictionary database is an annual report database that stores annual report information on corporate activities. The information processing system according to claim 2 .

4. Further, a search module is provided, The search module generates search keywords from the chat in the chat module, obtains second search results in a related information database to be searched, inputs the second search results into the large-scale language model, obtains the second search result answer, and displays it on the user terminal screen. The information processing system according to claim 1 .

5. a note module that executes a process of acquiring a summary of the chat in the chat module through the large-scale language model; a note database for storing the summary sentence; The chat module displays, on the screen of the user terminal, a note addition object for receiving an instruction to execute a process of saving the summary in the note database. The information processing system according to claim 1 .

6. the chat module displays, on a screen of the user terminal, a map generation object for receiving an instruction to execute a process for generating a visualized map that visualizes the chat in the chat module; When the process of generating the visualization map is executed, the visualization map is acquired through the large-scale language model and displayed on the screen of the user terminal. The information processing system according to claim 1 .

7. a proposal module that generates a proposal with content corresponding to the summary; a business plan database for storing the business plan; the note module displays, on the screen of the user terminal, a proposal generation object for receiving an instruction to execute a process for generating the proposal from the summary; The proposal module sets the status to either public, which allows viewing from a user terminal, or private, which does not allow viewing from the user terminal, and stores the status in the proposal database together with the proposal to be saved. The information processing system according to claim 5 .

8. Further, a search module is provided, The business plan database manages evaluations of the business plans, The search module displays the proposal objects linked to the proposal on the user terminal screen in order of the highest or lowest evaluation. The information processing system according to claim 7 .

9. Further, a search module is provided, The search module generates search keywords from the input prompt, obtains a third search result from at least the business plan database, inputs the third search result into the large-scale language model, obtains the answer, and displays it on the user terminal screen. The information processing system according to claim 7 .

10. An information processing method executed by an information processing system including a processor, a receiving step of receiving an input of a user prompt regarding a business idea; a chat step of inputting the user prompt into a large-scale language model, obtaining a response from the large-scale language model, and conducting a chat; The chat step includes: When a first user prompt is input, the first user prompt and a system prompt including logic for selecting one of a plurality of business frameworks in accordance with a wording included in the first user prompt are input to the large-scale language model; Obtain a first answer according to the selection framework selected in the large-scale language model and display it on a user terminal screen; As the answer to the second or subsequent user prompt, a second or subsequent answer in accordance with the selection framework is obtained and displayed on the user terminal screen. Information processing methods.

11. A program for executing the information processing method according to claim 10.

Images