Information processing system

Abstract

An information processing system according to the present invention comprises: a reception unit that receives input of a problem from a user; a first acquisition unit that acquires, from a database, a solving method for solving the problem received by the reception unit and an algorithm for realizing the solving method; a second acquisition unit that acquires, from the database, a prediction model which is suitable for the solving method acquired by the first acquisition unit; and a solution acquisition process unit that obtains a solution to the problem using the algorithm which has been acquired by the first acquisition unit and to which an execution condition for the algorithm has been applied and the prediction model acquired by the second acquisition unit.

Description

Information Processing Systems

[0001] The present technology relates to the technical field of information processing systems that obtain solutions to user problems.

[0002] Remarkable advances in AI (Artificial Intelligence) technology have led to the emergence of an AI model known as a Large Language Model (LLM). Large language models, such as those known as Generative Pre-trained Transformers (GPT) and Bidirectional Encoder Representations from Transformers (BERT), generate response information based on the content of an input string (prompt) and are capable of performing a variety of natural language processing tasks.

[0003] For example, when a question sentence is input as a prompt into the large-scale language model, answer information to the question is generated by performing a data search process according to the content of the question, a sentence generation process according to the search results, etc. In addition to the task of generating answer information to such a question, various tasks for generating response information according to the content of the input prompt can be executed, such as a task for creating a computer program that realizes the processing specified by the prompt, or a task for generating an image or music that satisfies the conditions specified by the prompt.

[0004] In the following Non-Patent Document 1, a large-scale language model is utilized to perform physical simulations.

[0005] Ruibo Liu, 7 others, "Mind's Eye: Grounded Language Model Reasoning through Simulation", (Submitted on 11 Oct 2022 (v1)), [Retrieved April 9, 2020], Internet <URL: https: / / doi.org / 10.48550 / arXiv.2210.05359>

[0006] Large-scale language models generate answers to questions by repeatedly selecting words that are likely to follow previously generated sentences. Therefore, answers provided by large-scale language models are not physically or mathematically guaranteed and may contain errors. Therefore, answers provided to specialized questions need to be carefully examined based on specialized knowledge, and there is room for improvement.

[0007] The present technology has been developed in consideration of such problems, and aims to present a physically or mathematically guaranteed solution.

[0008] The information processing system of the present technology includes a reception unit that receives a problem input from a user; a first acquisition unit that acquires, from a database, a solution for solving the problem received by the reception unit and an algorithm for implementing the solution; a second acquisition unit that acquires, from the database, a predictive model suitable for the solution acquired by the first acquisition unit; and a solution acquisition processing unit that acquires a solution to the problem using the algorithm to which execution conditions for the algorithm acquired by the first acquisition unit are applied and the predictive model acquired by the second acquisition unit. The reception unit, for example, has a function of interacting with a user as an AI agent. Furthermore, the "solution" acquired by the first acquisition unit is a method for solving the problem. For example, if the problem is to improve the characteristics of a product produced through multiple processes such as process A and process B, types of solutions include "changing parameter P1 in process A," "changing parameter P2 in process B," and "changing the processing order of each process." Furthermore, the "algorithm" acquired by the first acquisition unit is a method for implementing the solution. For example, "single-objective optimization," "multi-objective optimization," and "reinforcement learning," which are used to find an optimal solution for "parameter P1 in process A," are considered types of algorithms. For example, a solution obtained based on a response to a large-scale language model may not be a physically or mathematically guaranteed solution. However, according to this configuration, even when a large-scale language model is used, a solution is obtained by selecting a solution method or algorithm obtained from a database and then executing the algorithm.

[0009] 25 is a diagram showing a schematic configuration of an information processing system according to the present embodiment. FIG. 26 is a block diagram showing the configuration of a computer device. FIG. 27 is a functional block diagram of a server device. FIG. 28 is a diagram showing an example of information stored in a database. FIG. 29 is a diagram showing another example of information stored in a database. FIG. 29 is a block diagram explaining the functional configuration and data flow of a server device. FIG. 29 is a functional block diagram of a user terminal. FIG. 29 is a flowchart showing an example of processing executed by the CPU of the server device together with FIG. 9 and FIG. 10. FIG. 29 is a flowchart continuing from FIG. 8. FIG. 29 is a flowchart continuing from FIG. 9. FIG. 26 is a diagram showing a schematic configuration of an information processing system according to a second embodiment. FIG. 27 is a block diagram explaining the functional configuration and data flow of a server device according to the second embodiment. FIG. 28 is a diagram showing a schematic configuration of an information processing system according to a third embodiment. FIG. 29 is a block diagram explaining the functional configuration and data flow of a server device according to the third embodiment. FIG. 29 is a block diagram showing an example of the functional configuration of a server device. FIG. 29 is a diagram showing an example of a fishbone diagram, which is one aspect of a causal relationship diagram. FIG. 29 is a diagram showing another example of a fishbone diagram. FIG. 29 is a block diagram showing the functions of an information management unit in more detail. FIG. 29 is a diagram showing an example of graph data. FIG. 29 is a diagram showing an example of graph data in which all parent factors appear as nodes. FIG. 29 is a diagram showing a search range. FIG. 29 is a diagram showing an example of a first prompt. FIG. 29 is a diagram showing an example of a second prompt. FIG. 29 is a flowchart showing an example of processing executed by the CPU of the server device together with FIG. 25. Fig. 10 is a flowchart showing a processing flow according to a first modified example; Fig. 11 is a flowchart showing a processing flow according to a second modified example; Fig. 12 is a diagram showing an example of a fishbone diagram according to a modified example; Fig. 13 is a diagram showing another example of a fishbone diagram according to a modified example.

[0010] Hereinafter, with reference to the accompanying drawings, embodiments of an information processing device according to the present technology will be described in the following order: <1. Configuration of information processing system> <2. Example of hardware configuration of each device> <3. Example of functional configuration of each device> <3-1. Server device> <3-2. User terminal> <4. Processing flow> <5. Specific application examples> <5-1. Application example 1> <5-2. Application example 2> <5-3. Application example 3> <6. Second embodiment> <7. Third embodiment> <8. Modified example> <9. Database generation method> <10. Modified example> <11. Application example 4> <12. Summary> <13. This technology>

[0011] 1. Configuration of Information Processing System An information processing system 1A according to a first embodiment of the information processing system 1 of the present technology will be described with reference to the accompanying drawings.

[0012] 1, the information processing system 1A includes a server device 2A and a user terminal 3. A plurality of user terminals 3 may be provided in the information processing system 1A.

[0013] The server device 2A and the user terminal 3 are connected to a communication network NW such as the Internet, thereby enabling mutual communication.

[0014] The information processing system 1A is a system that generates and presents an answer to a question entered by a user on the user terminal 3. In the following explanation, the problem that the user wants to solve and that is expressed by the question will be referred to as "issue SJ." Furthermore, the solution corresponding to the issue SJ and that is expressed by an answer or the like presented by the information processing system 1A will be referred to as "solution AS."

[0015] The server device 2A grasps the input assignment SJ by communicating with the user terminal 3. The assignment SJ may be grasped by analyzing the acquired question. The server device 2A acquires or generates a solution AS corresponding to the assignment SJ and generates it as an answer to the assignment SJ. The server device 2A acquires information for obtaining the solution AS from a database DB under the management of the server device 2A.

[0016] The database DB stores past problems SJ and corresponding solutions AS, as well as the solution method MS, algorithm AR, and prediction model PM used to obtain the solutions AS, all of which are linked together.

[0017] The server device 2A generates or acquires a solution AS using information acquired from the database DB, and generates an answer sentence for presenting the solution AS. The server device 2A performs a process for displaying the answer sentence on a display unit or the like of the user terminal 3. To this end, the server device 2A appropriately executes, for example, a process for generating a web page or a process for generating a file. In the following description, the various processes executed by the server device 2A to present an answer to the user will be referred to as an "answer generation process."

[0018] The user terminal 3 performs processes of accepting input of the task SJ by the user, sending the input task SJ to the server device 2A, and receiving or displaying answer sentences, files, etc. from the server device 2A.

[0019] The server device 2A may receive input from the user as appropriate during the answer generation process. For example, the server device 2A may execute a process of confirming with the user whether or not the information in the database DB acquired for generating the answer is appropriate.

[0020] The server device 2A uses a generation AI (Artificial Intelligence), specifically, a Large Language Model (LLM), to set a task SJ based on a question sentence, obtain a solution AS, and then generate an answer sentence. More specifically, the server device 2A uses a large language model that interacts with a user in a chat format.

[0021] By using a large-scale language model, the server device 2A can advantageously perform the process of analyzing the input string (prompt) using natural language processing (NLP) technology to identify the task SJ, and the process of generating an answer sentence using natural language processing technology after obtaining a solution AS for the task SJ.

[0022] In the following description, the large-scale language model used by the server device 2A to interact with the user may be referred to as an "AI agent."

[0023] 2. Example of Hardware Configuration of Each Device An example of the specific hardware configuration of the server device 2A, the user terminal 3, and the database DB is shown in Fig. 2. When referring to the server device 2A, the user terminal 3, and the database DB without distinction, they will be referred to as "computer device Com."

[0024] As shown in the figure, the computer device Com includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, and a RAM (Random Access Memory) 13. The computer device Com may also include a processor other than the CPU 11, such as a GPU (Graphics Processing Unit).

[0025] The CPU 11 executes various processes according to programs stored in the ROM 12 or programs loaded into the RAM 13. The RAM 13 also stores data necessary for the CPU 11 to execute various processes as appropriate.

[0026] The functions realized by the CPU 11 of the server device 2A, the user terminal 3, and the database DB executing the programs will be described later.

[0027] The CPU 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output interface (I / F) 15 is also connected to the bus 14. An input unit 16 consisting of operators and operation devices is connected to the input / output interface 15. For example, the input unit 16 may be various operators and operation devices such as a keyboard, a mouse, keys, a dial, a touch panel, a touch pad, or a remote controller. A user operation is detected by the input unit 16, and a signal corresponding to the input operation is interpreted by the CPU 11.

[0028] A display unit 17, such as an LCD (Liquid Crystal Display) or an organic EL (Electro-Luminescence) panel, and an audio output unit 18, such as a speaker, are connected integrally or separately to the input / output interface 15. The display unit 17 is used to display various types of information, and may be, for example, a display device provided in the housing of the computer device, or a separate display device connected to the computer device.

[0029] The display unit 17 displays images for various image processing, moving images to be processed, etc. on the display screen based on instructions from the CPU 11. The display unit 17 also displays various operation menus, icons, messages, etc., i.e., a GUI (Graphical User Interface), based on instructions from the CPU 11.

[0030] For example, a GUI for a browser application or a dedicated application for receiving the problem-solving service provided by the server device 2A is displayed on the display unit 17 of the user terminal 3. In the user terminal 3, the input unit 16 detects an input operation on the GUI, and in accordance with the operation content interpreted by the CPU 11, processes such as displaying the input characters in a predetermined input field and transmitting the input content to the server device 2A are realized.

[0031] The input / output interface 15 may be connected to a storage unit 19 configured with a hard disk drive (HDD) or solid-state memory, or a communication unit 20 configured with a modem or the like.

[0032] In the information processing system 1A, the server device 2A and the database DB are provided as separate computers Com. However, the present invention is not limited to this, and the database DB may be provided as the storage unit 19 of the server device 2A.

[0033] The communication unit 20 performs communication processing via a transmission path such as the Internet, and communication with various devices via wired / wireless communication, bus communication, and the like.

[0034] A drive 21 is also connected to the input / output interface 15 as required, and a removable recording medium 22 such as a magnetic disk, optical disk, magneto-optical disk, or semiconductor memory is appropriately loaded therein.

[0035] The computer device Com can read data files such as programs used for various processes from a removable recording medium 22 using a drive 21. The read data files are stored in a storage unit 19, and images and sounds contained in the data files are output on a display unit 17 and an audio output unit 18. In addition, the computer programs and the like read from the removable recording medium 22 are installed in the storage unit 19 as needed.

[0036] In the computer device Com having the hardware configuration described above, for example, software for the processing of this embodiment can be installed via network communication by the communication unit 20 or via the removable recording medium 22. Alternatively, the software may be stored in advance in the ROM 12, the storage unit 19, etc. The CPU 11 performs processing operations based on various programs, thereby executing information processing and communication processing required for the server device 2A, the user terminal 3, or the database DB.

[0037] The server device 2A, the user terminal 3, or the database DB may not necessarily be configured by a single computer device Com as shown in Fig. 2, but may be configured by a system of multiple computer devices Com. The multiple computer devices Com may be systemized using a LAN (Local Area Network) or the like, or may be located in a remote location using a VPN (Virtual Private Network) or the like using the Internet. The multiple computer devices Com may include a computer device Com as a server group (cloud) available through a cloud computing service.

[0038] <3. Example of Functional Configuration of Each Device> <3-1. Server Device> FIG. 3 illustrates various functions that are realized as an AI agent by the CPU 11 serving as a processor of the server device 2A executing a program.

[0039] The CPU 11 of the server device 2A functions as a presentation processing unit F21, a reception unit F22, a first acquisition unit F23, a second acquisition unit F24, a solution acquisition processing unit F25, and a summary unit F26.

[0040] The presentation processing unit F21 performs processing to present various types of information to the user, for example, processing to present a web page containing various types of information. For example, the presentation processing unit F21 performs processing to present a confirmation page or the like on the display unit 17 of the user terminal 3 when the user is asked to confirm the solution AS for the assignment SJ in the process of obtaining the solution AS, and processing to present the solution AS, a summary based on the solution AS, or report information on the display unit 17 of the user terminal 3.

[0041] It should be noted that the presentation of information by the presentation processing unit F21 does not necessarily have to be in the form of a web page or the like displayed on the display unit 17 of the user terminal 3. For example, the presentation processing unit F21 may present to the user a document file compiling the solution AS, a summary, a report, or the like, a file generated by spreadsheet software, or a file for presentation.

[0042] The reception unit F22 receives input of the assignment SJ by the user using the user terminal 3. The input of the assignment SJ may be received in the form of a question. That is, the CPU 11 of the server device 2A may identify the assignment SJ by analyzing the question received by the reception unit F22 using natural language processing technology.

[0043] The first acquisition unit F23 acquires a solution MS and an algorithm AR from the database DB to obtain a solution AS for a task SJ set by a user or a task SJ identified by analyzing a question sentence.

[0044] Here, the solution method MS and the algorithm AR will be explained.

[0045] A "solution MS" is a method for solving a problem SJ. For example, if the problem SJ is to improve the characteristics of a product that is produced through multiple processes such as process A and process B, types of solutions MS would be "changing parameter P1 in process A," "changing parameter P2 in process B," or "changing the processing order of each process."

[0046] An "algorithm AR" is a method for realizing a solution method MS. For example, "execution of a calculation flow using a specific formula" used to find the optimal solution for "parameter P1 in process A," "single-objective optimization," "multi-objective optimization," and "reinforcement learning" are considered to be types of algorithm AR. Other types of algorithm AR include "dynamic programming," "heuristic methods," and "simulated annealing."

[0047] When acquiring the solution MS and the algorithm AR from the database DB, the first acquisition unit F23 uses the result of determining the similarity between the problem SJ to be solved and past problems SJ.

[0048] 4 shows an example of information stored in the database DB. The database DB stores previously solved problems SJ, and the solutions MS, algorithms AR, and prediction models PM used to solve the problems as response results at that time.

[0049] The database DB stores a plurality of records. A record is stored when a solution AS for a task SJ is generated or acquired, and is provided for each combination of the task SJ and the solution AS.

[0050] Specifically, each record is composed of a "record ID," which is an ID (identification) that can identify a single record, and is linked to a "task SJ," a "solution MS used," a "algorithm AR used," a "prediction model PM used," an "execution condition," and a "solution AS."

[0051] That is, one record links the "solution MS," "algorithm AR," and "prediction model PM" used to solve the "problem SJ," and further links the "execution conditions" of the algorithm AR with the resulting "solution AS."

[0052] It is not necessary for all information to be stored in one record stored in the database DB. For example, a record for a task SJ for which a solution AS can be obtained without using a prediction model PM does not need to be linked to a prediction model PM. Also, a record for a task SJ for which a solution AS can be obtained simply by solving a specific mathematical formula does not need to be linked to a solution method MS.

[0053] Furthermore, among the information stored in the database DB, for example, the prediction model PM, the prediction model PM itself may be stored in the database DB, or information on the location where the prediction model PM is stored (such as a URL (Uniform Resource Locator)) may be stored as link information.

[0054] The first acquisition unit F23 searches the database DB for records similar to the identified task SJ, and acquires information on the solution MS and algorithm AR associated with the records. The solution MS and algorithm AR acquired by the first acquisition unit F23 are presented to the user by the presentation processing unit F21.

[0055] The second acquisition unit F24 acquires from the database DB a prediction model PM used in an algorithm AR used to obtain a solution AS to a task SJ set by a user or a task SJ identified by analysis of a question sentence.

[0056] Specifically, the second acquisition unit F24 acquires the prediction model PM linked to the record identified by the first acquisition unit F23 from the database DB. The prediction model PM acquired by the second acquisition unit F24 is presented to the user by the presentation processing unit F21.

[0057] Here, the prediction model PM is used when the algorithm AR is executed. The prediction model PM is a model that outputs a predicted value as output data based on input data. The prediction model PM is preferably a model that calculates a predicted value by performing a physically or mathematically guaranteed operation, such as a surrogate model.

[0058] The solution acquisition processing unit F25 obtains a final solution AS corresponding to the task SJ using the solution method MS and algorithm AR selected by the first acquisition unit F23, and the prediction model PM selected by the second acquisition unit F24.

[0059] Specifically, the solution acquisition processor F25 uses a "single-objective optimization" algorithm AR to realize a certain solution MS. When a specific prediction model PM is used in the algorithm AR, the solution acquisition processor F25 generates multiple types of input data for the prediction model PM and inputs each input data into the prediction model PM to obtain multiple corresponding predicted values. The multiple input data input to the prediction model PM are data assigned within a range according to the execution conditions.

[0060] The prediction model PM itself returns predicted values ​​that are physically or mathematically guaranteed, but this does not apply if the input data is physically or mathematically inconsistent. In this configuration, the input data to the prediction model PM is assigned within an appropriate range according to the execution conditions, thereby eliminating physically or mathematically inconsistent input data, corrupted input data, etc. Therefore, the predicted values ​​output from the prediction model PM are more physically or mathematically guaranteed.

[0061] The solution acquisition processing unit F25 selects the predicted value most suitable for solving the problem SJ from the predicted values ​​output from the prediction model PM, for example, the predicted value that has the greatest improvement in performance, and generates a solution AS. The solution AS generated by the solution acquisition processing unit F25 may be the selected predicted value itself, or may be a numerical value obtained as a result of further calculation using the selected predicted value. The solution AS generated by the solution acquisition processing unit F25 is presented to the user by the presentation processing unit F21.

[0062] The solution acquisition processing unit F25 may generate an execution file that sequentially performs a process of repeatedly inputting input data to the prediction model PM and acquiring output data, and a process of selecting the optimal output data for solving the problem from the multiple output data. The execution file includes execution conditions.

[0063] The solution acquisition processing unit F25 uses information stored in the database DB when determining the execution conditions. An example of the information stored in the database DB is shown in FIG. 5. The database DB stores algorithms AR and their corresponding default setting files and execution programs in association with each other. The setting files and execution programs stored in the database DB are considered to be template information, so to speak.

[0064] The solution acquisition processing unit F25 acquires a setting file and an execution program for the determined algorithm AR from the database DB. Then, in order to adapt them to the current task SJ, the solution acquisition processing unit F25 determines execution conditions based on the determined solution method MS and algorithm AR, and modifies the setting file of the template based on the execution conditions. The solution acquisition processing unit F25 may also modify the execution program based on the execution conditions.

[0065] Here, the execution conditions based on the solution method MS and the algorithm AR include, for example, when "single-objective optimization" or "multi-objective optimization" is selected as the algorithm AR, various parameters, constraints, items and functions to be optimized, etc. When "reinforcement learning" is selected as the algorithm AR to be used, the execution conditions include parameters and reward functions.

[0066] The solution acquisition processing unit F25 creates an execution file to execute the execution program based on the setting file.

[0067] The execution file generated by the solution acquisition processing unit F25 may be presented to the user by the presentation processing unit F21. The presentation of the execution file to the user by the presentation processing unit F21 may be performed to allow the user to confirm the execution contents such as the prediction model PM, the conditions, and the range of input data.

[0068] By checking the contents of the executable file generated by the solution acquisition processing unit F25 and presented by the presentation processing unit F21, the user can confirm whether the range of input data for the prediction model PM is appropriate or whether other execution conditions are appropriate.

[0069] Information indicating the confirmation result for the executable file, that is, user evaluation information for the executable file, is transmitted from the user terminal 3 and is received by the receiving unit F22 of the server device 2A.

[0070] The solution acquisition processing unit F25 appropriately modifies the executable file based on the user's evaluation information and executes the executable file. The modified executable file may be provided to the user again to obtain a user evaluation. In other words, the generation of the executable file and the user's evaluation of it may be repeated multiple times.

[0071] The summarizing unit F26 generates a summary that summarizes the solution AS obtained by the solution acquisition processing unit F25. This processing is realized, for example, by using a natural language processing technique with a large-scale language model as a base model.

[0072] The solution AS generated by the solution acquisition processing unit F25 and the summary generated by the summarization unit F26 are appropriately presented to the user by the presentation processing unit F21.

[0073] An example of the functions and information flow of the CPU 11 of the server device 2A is shown in Figure 6. Note that the units further provided in the presentation processing unit F21, the reception unit F22, the first acquisition unit F23, the second acquisition unit F24, the solution acquisition processing unit F25, and the summarization unit F26 can be replaced with processes that realize the functions. For example, the assignment information identification unit F26a provided in the summarization unit F26 can be replaced with the assignment information identification process executed by the summarization unit F26.

[0074] An input sentence input by a user is received by an input sentence receiving section F22a of the receiving section F22.

[0075] If the input sentence is a request for solving the assignment SJ, the input sentence receiving unit F22a passes the input sentence to the assignment information identifying unit F26a of the summarizing unit F26.

[0076] The assignment information identification unit F26a identifies an assignment SJ from the input sentence and provides the assignment SJ to the first acquisition unit F23 and the second acquisition unit F24. The assignment information identification unit F26a may also acquire past similar response information from the database DB based on the identified assignment and provide it to the first acquisition unit F23 and the second acquisition unit F24 together with the information on the assignment SJ.

[0077] The first acquisition unit F23 acquires a solution MS by the solution acquisition unit F23a and an algorithm AR by the algorithm acquisition unit F23b based on the provided task SJ.

[0078] The acquired solution MS and algorithm AR are provided to a solution acquisition processing unit F25.

[0079] In addition, the second acquisition unit F24 acquires a prediction model PM by the prediction model acquisition unit F24a based on the provided assignment SJ.

[0080] The acquired prediction model PM is provided to a solution acquisition processing unit F25.

[0081] The solution acquisition processing unit F25 performs an acquisition process by the template acquisition unit F25a based on the provided algorithm AR, that is, a process of acquiring a template of a setting file or an execution program from the database DB.

[0082] Furthermore, the solution acquisition processing unit F25 executes an execution condition setting process by the execution condition setting unit F25b based on the provided solution MS and algorithm AR.

[0083] Furthermore, the solution acquisition processing unit F25 generates an execution file by the algorithm execution unit F25c and executes the execution file based on the provided prediction model PM and the set execution conditions.

[0084] Various results obtained by executing the executable file are provided to an information presentation unit F21a of the presentation processing unit F21 and presented to the user. In addition, the result information obtained by the solution acquisition processing unit F25 may be summarized by a presentation information summarization unit F26b of the summarization unit F26, as shown by the dashed line in Figure 6, and then presented to the user by the information presentation unit F21a.

[0085] <3-2. User Terminal> Next, various functions realized by the CPU 11 serving as a processor of the user terminal 3 executing a program will be illustrated in FIG.

[0086] The CPU 11 of the user terminal 3 functions as a UI processing unit F31, a receiving processing unit F32, and a transmitting processing unit F33.

[0087] The UI processing unit F31 receives, for example, a service provided by the processing of the server device 2A, in this example, a problem-solving service, and performs UI processing for issuing appropriate instructions to the server device 2A.

[0088] Specifically, the UI processing unit F31 performs processing to display software for receiving the problem-solving service on the display unit 17 of the user terminal 3. The screen displayed on the display unit 17 includes, for example, a chat field in which a response history with the AI ​​agent is displayed, an input field for posing questions to the AI ​​agent, a send button, and the like.

[0089] Furthermore, the UI processing unit F31 executes a corresponding process depending on the detection result of an operation on an operator such as the send button. For example, when it is detected that the send button has been pressed with a character string entered in an input field, the UI processing unit F31 displays the character string displayed in the input field in the chat field as the latest comment content, and also transmits the character string to the server device 2A.

[0090] The receiving processor F32 receives text information and files as response information from the server device 2A. The received text information is displayed in the chat box by the UI processor F31.

[0091] The transmission processing unit F33 transmits a character string indicating the content of the comment input by the user to the server device 2A.

[0092] The processes of the reception processing unit F32 and the transmission processing unit F33 are realized by the CPU 11 of the user terminal 3 controlling the communication unit 20.

[0093] 4. Process Flow The flow of various processes executed by the CPU 11 of the server device 2A will be described with reference to Figures 8, 9, and 10. Note that connections between processes in the figures are indicated by connectors C1, C2, and C3.

[0094] In step S101, the CPU 11 of the server device 2A accepts an input sentence input by the user via the user terminal 3.

[0095] In step S102, the CPU 11 of the server device 2A analyzes the input sentence and understands the content of the sentence. This process is performed using, for example, natural language processing technology.

[0096] In step S103, the CPU 11 of the server device 2A determines whether the content of the input sentence presents a task SJ, that is, whether it requests the solution of the task.

[0097] If it is determined that the content of the input sentence is not a request for problem solving (step S103: No), the CPU 11 of the server device 2A proceeds to step S104 and performs a corresponding process. The corresponding process differs depending on the analyzed content of the input sentence.

[0098] For example, if the content is a question about the meaning of a term, the CPU 11 of the server device 2A generates an explanation of the term using natural language processing technology and presents it to the user.

[0099] If it is determined that the content of the input statement is a request for problem solving (step S103: Yes determination), the CPU 11 of the server device 2A proceeds to step S105, where it acquires a solution MS and an algorithm AR for solving the problem SJ from the database DB. In this acquisition process, the CPU 11 of the server device 2A searches for similar problems SJ from past information, and acquires the solution MS and algorithm AR used to solve the corresponding problem SJ.

[0100] The solution MS and algorithm AR acquired by the CPU 11 of the server device 2A may be one or more. Furthermore, if the CPU 11 of the server device 2A does not find a similar task SJ, it may not need to acquire the solution MS and algorithm AR. In that case, the CPU 11 of the server device 2A may notify the user that a similar task SJ has not been found.

[0101] Here, the explanation will proceed assuming that a similar problem has been found.

[0102] In step S106, the CPU 11 of the server device 2A presents the solution MS and algorithm AR acquired from the database DB to the user. In this presentation process, the CPU 11 notifies the user of the contents of the solution MS and the algorithm AR, and also causes the user terminal 3 to present a user interface for the user to express their opinion as to whether the acquired solution MS and algorithm AR are appropriate.

[0103] For example, the CPU 11 of the server device 2A displays a sentence such as "May I proceed using the searched solution and algorithm?" on the display unit 17 of the user terminal 3, and accepts input of "Yes" or "No" from the user.

[0104] The CPU 11 of the server device 2A performs a process of accepting these user evaluations in step S107. This process can be said to be a process of receiving information on user evaluations.

[0105] In step S108, the CPU 11 of the server device 2A determines whether the user evaluation is a positive evaluation. If the user evaluation is a negative evaluation (step S108: No), the CPU 11 of the server device 2A returns to step S105 and again obtains the solution MS and the algorithm AR from the database DB.

[0106] If a negative evaluation is determined in step S108, the CPU 11 of the server device 2A may execute a process in which the user is asked to select an inappropriate solution MS or algorithm AR from the multiple solutions MS and algorithms AR presented to the user, and after eliminating the inappropriate solution MS or algorithm AR, proceed to the process of step S109.

[0107] If the CPU 11 of the server device 2A determines that the user evaluation is a positive evaluation (step S108: Yes judgment), it proceeds to step S109 and selects a solution MS and algorithm AR to be used to obtain a solution AS from one or more extracted solutions MS and algorithms AR.

[0108] Next, the CPU 11 of the server device 2A proceeds to the process of FIG.

[0109] In step S110, the CPU 11 of the server device 2A determines whether the solution method MS to be used uses the prediction model PM. If the prediction model PM is not linked to the solution method MS and algorithm AR acquired in step S109, the CPU 11 of the server device 2A determines "No" in step S110.

[0110] If it is determined that the solution MS to be used is a solution MS that does not require the prediction model PM (step S110: No determination), the CPU 11 of the server device 2A avoids the processing of step S111 and proceeds to step S112.

[0111] On the other hand, if it is determined that the solution MS to be used is a solution MS that uses a prediction model PM (step S110: Yes judgment), in step S111, the CPU 11 of the server device 2A obtains the prediction model PM linked to the solution MS and the algorithm AR from the database DB.

[0112] In step S112, the CPU 11 of the server device 2A sets execution conditions for the algorithm AR. The execution conditions may include the range of values ​​that can be taken by the input data input to the prediction model PM and other conditions. Examples of other conditions include the range of values ​​that can be taken by parameters that can be set for the prediction model PM and algorithm AR to be used.

[0113] In step S113, the CPU 11 of the server device 2A generates an executable file. As described above, this executable file includes a program that executes one or more times a process of inputting a predetermined range of input data to the selected prediction model PM and obtaining an output from the prediction model PM. The executable file may also include a process of selecting data that is optimal for solving the problem from one or more output data obtained from the prediction model PM.

[0114] In step S114, the CPU 11 of the server device 2A presents the execution file to the user. Note that, instead of presenting the execution file to the user, the CPU 11 of the server device 2A may present to the user information on the possible range of input data to the prediction model PM and execution conditions.

[0115] In step S115, the CPU 11 of the server device 2A receives the user's evaluation of the executable file. This process can be said to be a process of receiving information on the user's evaluation.

[0116] In step S116, the CPU 11 of the server device 2A determines whether the user evaluation regarding the executable file is a positive evaluation. If the user evaluation is a negative evaluation (step S116: No), the CPU 11 of the server device 2A returns to step S112 and sets the execution conditions again.

[0117] On the other hand, if it is determined that the user evaluation regarding the executable file is a positive evaluation (step S116: Yes determination), the CPU 11 of the server device 2A proceeds to step S117 and executes the created executable file.

[0118] Next, the CPU 11 of the server device 2A proceeds to the process of FIG.

[0119] In step S118, the CPU 11 of the server device 2A presents the execution result of the algorithm AR to the user. Note that the information presented to the user here is information before the solution AS is obtained, such as the output results of one or more prediction models PM according to the number of input data. However, the CPU 11 of the server device 2A may also present the final solution AS in step S118.

[0120] In step S119, the CPU 11 of the server device 2A performs a process of accepting a user evaluation of the presented execution result. This process can be said to be a process of receiving information on the user evaluation.

[0121] In step S120, the CPU 11 of the server device 2A determines whether the user evaluation regarding the execution result is a positive evaluation. If the user evaluation is a negative evaluation (step S120: No), the CPU 11 of the server device 2A returns to step S112 in FIG. 9 and starts over from setting the execution conditions.

[0122] On the other hand, if it is determined that the user evaluation of the executable file is a positive evaluation (step S120: Yes judgment), the CPU 11 of the server device 2A proceeds to step S121, generates a solution AS from the obtained execution result, and presents it to the user.

[0123] In step S122, the CPU 11 of the server device 2A generates a report summarizing the details of the generated solution AS. The report may be generated by using the function of the summary section F26 to generate a summary to be displayed in the chat box, or may be generated as a document file, a spreadsheet file, a presentation file, or the like.

[0124] The CPU 11 of the server device 2A presents the report in step S123, and ends the series of processes shown in Figures 8, 9, and 10. Note that the CPU 11 of the server device 2A may return to the process of step S101 after step S123.

[0125] The processes shown in FIGS. 8, 9, and 10 can be considered as the answer generation process described above.

[0126] 5. Specific Application Examples Several examples of input of a task SJ by a user and presentation of various information by the server device 2A in response to the input will be shown.

[0127] <5-1. Application Example 1> The first application example is an example of solving problem SJ related to the design of pixels provided in an image sensor. Specifically, this is an example of a case where a user inputs a sentence such as, "For a new image sensor equipped with pixels that employ design type AA, please tell me a method for improving the saturation signal amount by 20% or more while relaxing the design margin for the dark signal characteristics of the image."

[0128] This input sentence is received by the input sentence receiving unit F22a shown in FIG. 6, and the assignment SJ is identified by the assignment information identifying unit F26a.

[0129] The assignment information identification unit F26a or the first acquisition unit F23 searches the database DB based on the identified assignment, and identifies, for example, the following information D1, D2, and D3 as available information.

[0130] Information D1: Design materials and result information regarding the saturated signal amount in design type AA. Information D2: Information of record ID=id1 in which the study materials and improvement results of design type AB, which has a pixel size similar to that of design type AA, are linked. Information D3: Information of record ID=id2 in which the study materials and improvement results of design type AC, which has a manufacturing flow similar to that of design type AA, are linked.

[0131] Here, the information D1 is information different from past response information, and is result information that indicates a change in performance when parameters are changed in product design documents, simulations, or actual equipment. The information D1 may be stored in the database DB or in a device separate from the database DB.

[0132] These search results acquired by the assignment information identification unit F26a and the presentation processing unit F21 may be presented to the user by the presentation processing unit F21 as appropriate.

[0133] The solution acquisition unit F23a identifies the solution MS used for each of the acquired information D1, D2, and D3, and acquires information about the solution MS.

[0134] For example, the following three solution MSs are obtained.

[0135] Solution MS information associated with information D1: Optimization of the size of the photomask that affects the implantation area in the ion implantation process Solution MS information associated with information D2: Optimization of the impurity concentration in the ion implantation process to improve the saturation signal amount by 20% Solution MS information associated with information D3: Optimization of the impurity concentration in the ion implantation process to improve the saturation signal amount by 20%

[0136] The algorithm acquisition unit F23b identifies the algorithm AR used for each of the acquired information D1, D2, and D3, and acquires information about the algorithm AR. An example of the acquired algorithm AR is shown below.

[0137] Information on the algorithm AR associated with information D1: Maximizing the amount of saturated signals using reinforcement learning Information on the algorithm AR associated with information D2: Maximizing the amount of saturated signals using reinforcement learning Information on the algorithm AR associated with information D3: Maximizing the amount of saturated signals using single-objective optimization

[0138] The information presenting unit F21a presents the information acquired by the solution acquiring unit F23a and the algorithm acquiring unit F23b to the user in the following form, for example.

[0139] "In the past information for design type AA, the saturated signal level was improved by optimizing the photomask. In the past information for design types AB and AC, the saturated signal level was improved by optimizing the impurity concentration in the ion implantation process. In addition, in the past information for design types AA and AB, optimization was performed using reinforcement learning. In addition, in the past information for design type AC, optimization was performed using single-objective optimization."

[0140] Then, the information presenting unit F21a performs a process of requesting the user to input an evaluation by presenting the following sentence:

[0141] "Based on the above information, we will use reinforcement learning to optimize the impurity concentration in the ion implantation process and present the results of our study. Is it okay to proceed?"

[0142] The user gives a positive or negative evaluation of the proposed method by inputting "Yes" or "No" to this suggestion statement.

[0143] In response to this, for example, if the user inputs "Yes," the CPU 11 of the server device 2A determines to use the solution method MS of "optimization of impurity concentration in an ion implantation process" and the algorithm AR of "reinforcement learning." With this determination, the server device 2A determines to obtain the solution AS by referring to the information D2.

[0144] Next, the prediction model acquisition unit F24a acquires a prediction model PM based on the information D2. The acquired prediction model PM is referred to as a prediction model PM1.

[0145] Next, the template acquisition unit F25a acquires the algorithm AR used to solve the current task SJ, that is, a template of a setting file and an execution program for performing reinforcement learning, from the database DB.

[0146] Next, the execution condition setting unit F25b sets the execution conditions, which include, for example, information such as parameters to be adjusted, reward functions, desired performance information, and design margins, and are set so that these fall within appropriate physical or mathematical ranges.

[0147] In this example, the parameter to be adjusted is, for example, the impurity concentration itself or a parameter that affects the impurity concentration. The reward function is the reward function used in information D2. The performance information to be realized is the value obtained by multiplying the design value of the saturation signal amount by 1.2, i.e., a value indicating a 20% improvement in the saturation signal amount. The design margin is, for example, a parameter indicating the design margin of the dark signal characteristics of the image.

[0148] The algorithm execution unit F25c generates an execution file taking into account the execution conditions.

[0149] Then, the information presenting unit F21a performs a process of requesting the user to input an evaluation by presenting the executable file and the following sentence.

[0150] "Are you sure you want to proceed with the design under these execution conditions?"

[0151] The user gives a positive or negative evaluation of the execution conditions, etc. by inputting "Yes" or "No" to this suggestion statement.

[0152] In response to this, for example, if the user inputs "Yes", the algorithm execution unit F25c of the server device 2A executes the created executable file and obtains the result.

[0153] On the other hand, instead of inputting "No" to the above-mentioned suggested sentence, the user can change the constraint conditions such as parameters. For example, the user can input the following input sentence:

[0154] "Please relax the design margin for dark signal characteristics by 50%."

[0155] When such an input is made, the execution condition setting unit F25b reconsiders the execution conditions and presents them to the user again.

[0156] The information presentation unit F21a performs processing to compile the execution results of the algorithm execution unit F25c into text, images, or files and present them to the user. Note that the presentation of the result information by the algorithm execution unit F25c may be performed by presenting a summary obtained by summarizing processing performed by the presentation information summarization unit F26b.

[0157] For example, the user may be presented with the following sentence as a summary:

[0158] "As a result of performing optimization, we were able to improve the saturation signal amount by 30% by setting the impurity concentration in the ion implantation process to X1%. The design margin for the dark signal characteristics at this time is Y1%."

[0159] In addition, the file that the information presentation unit F21a presents to the user may be, for example, a file that describes various data for each input of input data to the prediction model PM that is repeatedly executed and each acquisition of output data, such as a file that describes information such as the design margins and reward functions for the saturation signal amount and dark signal characteristics in correspondence with each other.

[0160] 5-2. Application Example 2 The second application example is an example of solving problem SJ related to the design of a micro-channel chip in which a micro-channel structure is formed on a substrate by microfabrication. Specifically, this is an example of a case in which a user inputs the following sentence: "For the design of micro-channel chip type BB, please propose a design plan that maximizes the flow rate in section CD under manufacturable conditions."

[0161] The assignment information identification unit F26a or the first acquisition unit F23 searches the database DB based on the identified assignment, and identifies, for example, the following information D4 and D5 as usable information.

[0162] Information D4: Design materials and result information regarding the conditions for manufacturing in design type BB. Information D5: Record ID=id3 information linked to the study materials and improvement results of the old design type BC with the same structure as design type BB.

[0163] Here, the information D4 is information different from past response information, and is result information that indicates a change in performance when parameters are changed in product design documents, simulations, or an actual device. The information D4 may be stored in the database DB or in a database separate from the database DB.

[0164] These search results acquired by the assignment information identification unit F26a and the presentation processing unit F21 may be presented to the user by the presentation processing unit F21 as appropriate.

[0165] The solution acquisition unit F23a identifies the solution MS used for each of the acquired information D4 and D5, and acquires information about the solution MS.

[0166] For example, the following two solutions MS are acquired: Note that although the solution MS is referred to as the solution MS, the information on the solution MS linked to the information D4 can be rephrased as information on the conditions under which manufacturing is possible.

[0167] Solution MS information associated with information D4: Set the conditions so that the cross-sectional area is X2 or more and the length of each side is Y2 or more. Solution MS information associated with information D5: Adjust the cross-sectional area using information D5 and general information.

[0168] The algorithm acquisition unit F23b identifies the algorithm AR used for each of the acquired information D4 and D5, and acquires information about the algorithm AR. Note that information D4 is information about manufacturable conditions, and since no algorithm AR exists for information D4, an example of the algorithm AR acquired for information D5 is shown here.

[0169] Algorithm AR information linked to information D5: Analyzing the trade-off between flow velocity and pressure loss using multi-objective optimization

[0170] The information presenting unit F21a presents the information acquired by the solution acquiring unit F23a and the algorithm acquiring unit F23b to the user in the following form, for example.

[0171] "According to the past information for design type BB, the cross-sectional area must be at least X2, and the length of each side must be at least Y2. In the past information for design type BC, the cross-sectional area was adjusted using information D5 and general information. In addition, in the past information for design type BC, multi-objective optimization was used to analyze the trade-off between flow velocity and pressure loss."

[0172] Then, the information presenting unit F21a performs a process of requesting the user to input an evaluation by presenting the following sentence:

[0173] "Based on the above information, we will use multi-objective optimization to adjust the cross-sectional area within the constraints and present the results of our study. Is it okay to proceed?"

[0174] The user gives a positive or negative evaluation of the proposed method by inputting "Yes" or "No" to this suggestion statement.

[0175] In contrast, if the user inputs "Yes," for example, the CPU 11 of the server device 2A determines to use the solution method MS of "adjusting the cross-sectional area within the constraints" and the algorithm AR of "multi-objective optimization." With this determination, the server device 2A determines to obtain the solution AS by referring to the information D4 and information D5.

[0176] Next, the prediction model acquisition unit F24a acquires a prediction model PM based on the information D5. The acquired prediction model PM is referred to as a prediction model PM2.

[0177] Next, the template acquisition unit F25a acquires the algorithm AR used to solve the current problem SJ, that is, a template of a setting file and an execution program for performing multi-objective optimization, from the database DB.

[0178] Next, the execution condition setting unit F25b sets the execution conditions, which include, for example, information such as parameters to be adjusted, optimization indices, and design margins, and are set so that these fall within appropriate physical or mathematical ranges.

[0179] In this example, the parameters to be adjusted are, for example, the cross-sectional area itself or parameters that affect the cross-sectional area. The optimization indices are the flow velocity and pressure loss. The design margins are, for example, parameters that indicate the cross-sectional area or the length of each side.

[0180] The algorithm execution unit F25c generates an execution file taking into account the execution conditions.

[0181] Then, the information presenting unit F21a performs a process of requesting the user to input an evaluation by presenting the executable file and the following sentence.

[0182] "Are you sure you want to proceed with the design under these execution conditions?"

[0183] The user gives a positive or negative evaluation of the execution conditions, etc. by inputting "Yes" or "No" to this suggestion statement.

[0184] In response to this, for example, if the user inputs "Yes", the algorithm execution unit F25c of the server device 2A executes the created executable file and obtains the result.

[0185] On the other hand, instead of inputting "No" to the presented sentence, the user can change the constraint conditions such as parameters. In this case, the execution condition setting unit F25b reconsiders the execution conditions based on the analysis results of the input sentence and presents them to the user again.

[0186] The information presentation unit F21a performs processing to compile the execution results of the algorithm execution unit F25c into text, an image, or a file and present it to the user. The image may be, for example, a graph image showing the relationship between pressure loss on the X axis and flow velocity on the Y axis. The presentation of the result information by the algorithm execution unit F25c may be performed by presenting a summary obtained by summarizing processing performed by the presentation information summarization unit F26b.

[0187] For example, the user may be presented with the following sentence as a summary:

[0188] "After running the optimization, the length of each side when the flow velocity is at its maximum is Y3. The pressure loss at that time is Z2."

[0189] 5-3. Application Example 3 The third application example is an example of solving a problem SJ related to the design of pixels provided in an image sensor. Specifically, this is an example of a case where a user inputs a sentence such as, "For a new image sensor equipped with pixels that employ design type CC, please tell me the conditions that maximize the yield based on prototype data DD."

[0190] This input sentence is received by the input sentence receiving unit F22a shown in FIG. 6, and the assignment SJ is identified by the assignment information identifying unit F26a.

[0191] The assignment information identification unit F26a or the first acquisition unit F23 searches the database DB based on the identified assignment, and identifies, for example, the following information D6 as available information.

[0192] Information D6: Materials examining the conditions for maximizing yield based on prototype data DE in design type CC

[0193] These search results acquired by the assignment information identification unit F26a and the presentation processing unit F21 may be presented to the user by the presentation processing unit F21 as appropriate.

[0194] The solution acquisition unit F23a identifies the solution MS used in the acquired information D6 and acquires information about the solution MS. However, in this example, the solution MS used in the information D6 does not exist. Therefore, the solution acquisition unit F23a cannot identify the solution MS used in the information D6 and cannot acquire the information. That is, in step S105 of FIG. 8, the CPU 11 of the server device 2A may acquire only the algorithm AR without acquiring the solution MS. Similarly, in step S106, the CPU 11 of the server device 2A may present only the algorithm AR without presenting the solution MS.

[0195] The algorithm acquisition unit F23b identifies the algorithm AR used in the acquired information D6 and acquires information about the algorithm AR. An example of the acquired algorithm AR is shown below.

[0196] Information of the algorithm AR linked to the information D6: Using the formula (1) for calculating the yield, the variable that maximizes the yield is found.

[0197] The information presenting unit F21a presents the information acquired by the solution acquiring unit F23a and the algorithm acquiring unit F23b to the user in the following form, for example.

[0198] "There is no solution that can be presented. In design type CC, we use formula (1) to calculate yield to find the conditions that maximize yield."

[0199] Then, the information presenting unit F21a performs a process of requesting the user to input an evaluation by presenting the following sentence:

[0200] "Based on the above information, we will apply formula (1) to the prototype data DD and calculate the conditions that maximize the yield. Is it okay to proceed?"

[0201] The user gives a positive or negative evaluation of the proposed method by inputting "Yes" or "No" to this suggestion statement.

[0202] In response to this, for example, if the user inputs "Yes," the CPU 11 of the server device 2A determines to use the algorithm AR "executing a calculation flow using a specific formula (1)." With this determination, the server device 2A determines to obtain the solution AS by referring to the information D6.

[0203] Next, the prediction model acquisition unit F24a acquires a prediction model PM based on the information D6. However, in this example, experimental data is used, so there is no need to use the prediction model PM. Therefore, the prediction model acquisition unit F24a does not acquire the prediction model PM. That is, the CPU 11 of the server device 2A determines "No" in step S110 of FIG. 9.

[0204] Next, the template acquisition unit F25a acquires from the database DB an algorithm AR to be used to solve the current problem SJ, i.e., a template of a program for "executing a calculation flow using a specific formula (1)." This program is a program that reads the prototype data DD and calculates the yield.

[0205] Next, the execution condition setting unit F25b sets the execution conditions, which may include, for example, the data to be read and information on the mathematical formula to be applied.

[0206] In this example, the data to be read is the prototype data DD, and the formula to be applied is formula (1).

[0207] The algorithm execution unit F25c generates an execution file taking into account the execution conditions.

[0208] Then, the information presenting unit F21a performs a process of requesting the user to input an evaluation by presenting the executable file and the following sentence.

[0209] "Are you sure you want to proceed with the design under these execution conditions?"

[0210] The user gives a positive or negative evaluation of the execution conditions, etc. by inputting "Yes" or "No" to this suggestion statement.

[0211] In response to this, for example, if the user inputs "Yes", the algorithm execution unit F25c of the server device 2A executes the created executable file and obtains the result.

[0212] On the other hand, instead of inputting "No" to the presented sentence, the user can change the formula or the prototype data to be read in. In this case, the execution condition setting unit F25b reconsiders the execution conditions based on the analysis results of the input sentence and presents them to the user again.

[0213] The information presentation unit F21a performs processing to compile the execution results of the algorithm execution unit F25c into text, images, or files and present them to the user. Examples of images include bar graph images showing the yield for each of the conditions ZZ1, ZZ2, and ZZ3. The presentation of the result information by the algorithm execution unit F25c may be performed by presenting a summary obtained by summarizing the results using the presentation information summary unit F26b.

[0214] For example, the user may be presented with the following sentence as a summary:

[0215] "The condition that maximizes yield is condition ZZ3."

[0216] 6. Second Embodiment An information processing system 1B according to a second embodiment is an example in which the function of the solution acquisition processing unit F25 is executed by a device different from the server device 2.

[0217] FIG. 11 shows an example of the configuration of the information processing system 1B.

[0218] The information processing system 1B includes an application server device 4 in addition to a server device 2B, a user terminal 3, and a database DB managed by the server device 2B.

[0219] 12 is a diagram showing an example of the functions and information flow of the CPU 11 of the server device 2B and the application server device 4. The server device 2B functions as the presentation processing unit F21, the reception unit F22, the first acquisition unit F23, the solution acquisition processing unit F25, and the summarization unit F26 described above by the CPU 11 executing a program.

[0220] Furthermore, the server device 2B functions as a solution acquisition processing unit F25B when the CPU 11 executes a program.

[0221] By using the functions provided by the application server device 4, the solution acquisition processing unit F25B realizes the functions of the prediction model acquisition unit F24a that the second acquisition unit F24 of the server device 2A had, the functions of the template acquisition unit F25a that the solution acquisition processing unit F25 of the server device 2A had, the functions of the execution condition setting unit F25b, and the functions of the algorithm execution unit F25c.

[0222] Specifically, the application server device 4 has a function as a prediction model acquisition unit F24a, a function as a template acquisition unit F25a, a function as an execution condition setting unit F25b, and a function as an algorithm execution unit F25c.

[0223] That is, when the problem SJ, solution MS, and algorithm AR are input, the application server device 4 identifies the prediction model PM to be used in the algorithm AR, obtains a template based on the solution MS and the problem SJ, sets execution conditions, and then executes the algorithm AR. The application server device 4 also provides the execution result of the algorithm AR to the server device 2B.

[0224] Therefore, the solution acquisition processing unit F25B of the server device 2B only needs to generate input data for the application server device 4, thereby reducing the processing load on the server device 2B.

[0225] The information processing system 1B may use an algorithm execution tool installed in the server device 2B to realize the functions of the prediction model acquisition unit F24a, the template acquisition unit F25a, the execution condition setting unit F25b, and the algorithm execution unit F25c, instead of including the application server device 4. That is, the solution acquisition processing unit F25B of the server device 2B may include the functions of the application server device 4.

[0226] The application server device 4 and the algorithm execution tool may be prepared by the company that manages the information processing system 1C, or may be provided by, for example, an EDA (Electronic Design Automation) vendor.

[0227] 7. Third Embodiment The third embodiment is an example in which the server device 2 is configured by a plurality of devices.

[0228] An example is shown in FIG.

[0229] The information processing system 1C includes a plurality of server devices 2C1 and 2C2, a database DB managed by the server device 2C2, and a user terminal 3.

[0230] The server device 2C1, the server device 2C2, and the user terminal 3 are connected to each other via a communication network NW so as to be able to communicate with each other.

[0231] The server device 2C1 is an information processing device that transmits and receives information to and from the user terminal 3. A user uses the user terminal 3 to transmit various requests to the server device 2C1, thereby enjoying the services provided by the information processing system 1C.

[0232] Server device 2C1 transmits a request to server device 2C2 as appropriate in response to a user request received via user terminal 3. That is, server device 2C1 realizes the user request by utilizing server device 2C2.

[0233] Specifically, FIG. 14 shows the functions and information flow of the server device 2C1 and the server device 2C2.

[0234] For example, the server device 2C1 is an information processing device having the functions of a reception unit F22 that receives text input from a user, a summary unit F26 that performs a process of summarizing the received text and a process of summarizing information to be presented to the user, and a presentation processing unit F21 that presents information to the user.

[0235] The server device 2C2 is an information processing device that performs appropriate processing in response to a request received from the server device 2C1. As shown in Fig. 14, the server device 2C2 is an information processing device that has the functions of a first acquisition unit F23 that acquires information on the solution method MS and the algorithm AR from the database DB, a second acquisition unit F24 that acquires information on the prediction model PM, and a solution acquisition processing unit F25 that acquires a solution AS for the task SJ based on the information.

[0236] Although FIG. 13 shows an example in which the database DB is provided outside the server device 2C2, the database DB may be provided inside the server device 2C2 as shown in FIG.

[0237] The server device 2C1 analyzes the input sentence from the user, and if the problem SJ is identified, the server device 2C1 requests the server device 2C2 to generate a solution AS for solving the problem SJ.

[0238] Server device 2C2 receives the request to generate a solution AS, performs each process, and outputs the obtained solution AS to server device 2C1.

[0239] In the server device 2C1, the presentation processing unit F21 receives information on the solution AS and information on the report from the server device 2C2. The information presentation unit F21a of the presentation processing unit F21 may summarize the information to be presented to the user using the function of the presentation information summary unit F26b of the summary unit F26 as appropriate.

[0240] That is, the processing of the reception unit F22 is executed by a first processor provided in the server-side device of the information processing system 1C, i.e., the CPU 11 of the server device 2C1. Also, the processing of the first acquisition unit F23 is executed by a second processor provided in the server-side device of the information processing system 1C, i.e., the CPU 11 of the server device 2C2. In this way, the processing of the reception unit F22 and the first acquisition unit F23 may be executed by different processors, thereby distributing the load.

[0241] The processing of the second acquisition unit F24 is executed by a second processor included in the server-side device of the information processing system 1C, i.e., the CPU 11 of the server device 2C2. In this way, the processing of the reception unit F22 and the second acquisition unit F24 may be executed by different processors, thereby distributing the load.

[0242] Furthermore, the processing of the solution acquisition processing unit F25 is executed by a second processor provided in the server-side device of the information processing system 1C, i.e., the CPU 11 of the server device 2C2. In this way, the processing of the reception unit F22 and the processing of the solution acquisition processing unit F25 may be executed by different processors, thereby distributing the load.

[0243] 8. Modifications In the third embodiment, a configuration has been described in which the functions of the presentation processing unit F21, the reception unit F22, the first acquisition unit F23, the second acquisition unit F24, the solution acquisition processing unit F25, and the summarization unit F26 are realized using two server devices 2C1 and 2C2. This is not limiting, and each function may be realized using three or more server devices 2. For example, a server device 2 having the function of the first acquisition unit F23, a server device 2 having the function of the second acquisition unit F24, and a server device 2 having the function of the solution acquisition processing unit F25 may each be provided as separate information processing devices.

[0244] 9. Database Generation Method The database (DB) described above may be, for example, a graph database represented by information and distance information between each piece of information. In the graph database, each piece of information is represented as a "node (ND)," and the connections between pieces of information are represented as "edges (ED)."

[0245] The distance DT of each piece of information can be expressed as the number of edges ED included when the nodes ND are connected by the shortest distance. In other words, the distance DT between two pieces of information connected by one edge ED is set to "1".

[0246] Here, we will describe a method for constructing a database DB that is a graph database.

[0247] First, the database DB stores causal relationship diagrams that are presented to the user as appropriate in response to the input of a problem-solving prompt. Causal relationship diagrams can include various types of diagrams that are expressed by connecting information with lines, such as fishbone diagrams (CE diagrams) and mind maps. In the following explanation, a fishbone diagram will be used as an example.

[0248] The database DB construction function may be provided, for example, in the server device 2A (2B, 2C) in the information processing system 1 (1A, 1B, 1C) described above, or in the database DB calculation unit. In the following description, the server devices 2A, 2B, and 2C will be referred to as the "server device 2" without distinction.

[0249] FIG. 15 shows an example in which the server device 2 has a function of constructing a database DB.

[0250] The CPU 11 of the server device 2 functions as an information management unit F27 in addition to a presentation processing unit F21, a reception unit F22, a first acquisition unit F23, a second acquisition unit F24, a solution acquisition processing unit F25, and a summary unit F26.

[0251] The information management unit F27 performs a process of adding new information to the database DB and reconstructing it. Note that the information management unit F27 may also have a function of acquiring information from the database DB as described above. In other words, the information management unit F27 may have a function of executing various processes related to the information stored in the database DB.

[0252] For example, in response to a user's prompt, "Please analyze and explain the cause of the network failure," the information management unit F27 retrieves one fishbone diagram from multiple fishbone diagrams stored in the database DB. The retrieved fishbone diagram is presented to the user by the presentation processing unit F21.

[0253] An example of a fishbone diagram presented to the user is shown in Fig. 16. In the fishbone diagram shown in Fig. 16, a problem SJ called "occurrence of a network failure" is placed, and multiple parent factors are placed on the spine connected to the problem SJ.

[0254] The parent factors in the fishbone diagram shown in FIG. 16 are "people," "machine," "method," "material," and "measurement."

[0255] Furthermore, the child factors of the parent factor "people" are "misconfiguration," "lack of training," "operational error," "misconfiguration of permissions," and "lack of communication."

[0256] In the example shown in FIG. 16, there is no grandchild factor connected to a child factor, but a grandchild factor may be connected to a child factor.

[0257] For the presented fishbone diagram, the user decides and provides feedback on whether or not to have the fishbone diagram used to present a solution AS to the problem SJ, i.e., whether or not to use the presented fishbone diagram to solve the problem SJ.

[0258] If the presented fishbone diagram is rejected by the user, the information management unit F27 acquires a different fishbone diagram from the database DB. The acquired fishbone diagram is presented again by the presentation processing unit F21.

[0259] An example of the reacquired fishbone diagram is shown in FIG.

[0260] As shown in the figure, the reacquired fishbone diagram differs from the initially acquired fishbone diagram in the parts indicated by the thick dashed rectangles.

[0261] For example, in the fishbone diagram shown in FIG. 16, the child factors connected to the parent factor "method" are "GUI (Graphical User Interface) setting error," "DHCP (Dynamic Host Configuration Protocol) setting error," "firewall," "protocol setting error," and "incorrect network sharing setting."

[0262] On the other hand, in the fishbone diagram shown in FIG. 17, the child factors connected to the same parent factor "method" are "CLI (Command Line Interface) setting error," "Network service failure," "Routing setting error," "DNS (Domain Name System) setting error," and "Packet filter misconfiguration."

[0263] In the fishbone diagram shown in FIG. 17, "IP address misconfiguration" is connected as a grandchild cause to the child cause "Network service failure."

[0264] The fishbone diagram shown in FIG. 16 is for solving a network failure in a server using, for example, Company A's OS (Operating System).

[0265] On the other hand, the fishbone diagram shown in FIG. 17 is for solving a network failure in a server using an open source OS.

[0266] In this way, the fishbone diagram selected varies depending on the individual situation and conditions of the user. A plurality of types of fishbone diagrams can be stored in the database DB for the same type of task SJ.

[0267] When the user indicates his / her intention to adopt the presented fishbone diagram, the intention information is transmitted to the server device 2 .

[0268] The server device 2 receives the intention information through the reception unit F22, and the information management unit F27 performs processing to generate information to be presented to the user next.

[0269] The functions of the information management unit F27 are shown in Fig. 18. The information management unit F27 has a function as a search processing unit F27a and a function as a graph data generation unit F27b that generates information based on graph theory (hereinafter referred to as "graph data GD").

[0270] The graph data generation unit F27b generates graph data GD having nodes ND and distances DT based on the fishbone diagram selected by the user.

[0271] To this end, the graph data generation unit F27b first performs processing to output text information using a fishbone diagram as an image as input. This processing is performed using, for example, an AI model that extracts text information from an image.

[0272] An example of text information that is output when the fishbone diagram shown in FIG. 17 is input is shown below.

[0273] "Network failure" = {node1: person, node2: configuration error, insufficient training, operational error, permission setting error, lack of communication}, {node1: machine, node2: network equipment, hardware failure, network card, server failure, driver failure}, {node1: method, node2: CLI configuration error, {network service error, node3: IP address misconfiguration}, routing configuration error, DNS configuration error, packet filter misconfiguration}, {node1: materials, node2: faulty cable, faulty wiring, incorrect cable, inappropriate hardware, server configuration error}, {node1: measurement, node2: poor log acquisition, monitoring tool, inadequate server log, faulty measuring equipment, inaccurate data}

[0274] The text information generated in this manner includes information on the nodes ND and the distances DT between the nodes ND. The graph data generation unit F27b generates graph data GD using the text information. An example of the generated graph data GD is shown in FIG. 19.

[0275] The graph data GD is data in which child factors and grandchild factors are arranged according to the distance DT from a parent factor "method" based on the fishbone diagram shown in FIG.

[0276] The graph data generation unit F27b further associates information (such as the information D1 and information D2 described above) with each of the child factors and grandchild factors in the graph data GD.

[0277] For example, information linked to child factors and grandchild factors is stored in a database DB. The search processing unit F27a extracts information that has a high affinity with each factor from the information stored in the database DB. The graph data generation unit F27b links the information extracted by the search processing unit F27a to each factor. Here, this information is referred to as "materials."

[0278] For example, for the sub-cause "CLI setting error," a collection of case studies of when CLI setting errors occur, as well as documents summarizing countermeasures and solutions, can be linked.

[0279] In the example shown in FIG. 19, each of materials A to M is linked to one of the child factors or grandchild factors.

[0280] The graph data GD thus generated has a different data amount for each child factor and each grandchild factor. The data amount is expressed, for example, in bytes.

[0281] The graph data GD is stored in a database DB as appropriate.

[0282] Note that Figure 19 only shows the graph data GD for one parent factor "method" for the problem SJ "occurrence of a network failure", but in reality, for the problem SJ "occurrence of a network failure", graph data GD for the parent factor "people", graph data GD for the parent factor "machine", graph data GD for the parent factor "materials", graph data GD for the parent factor "measurement", etc. are stored in the database DB.

[0283] Furthermore, instead of generating graph data GD for each parent factor, one graph data GD may be stored for one task SJ. Specifically, the task SJ may also be treated as one node ND, and graph data GD in which each parent factor is connected to the node ND as the task SJ by an edge ED may be stored in the database DB (see FIG. 20). Note that in FIG. 20, the linking of materials is omitted from the illustration.

[0284] The first acquisition unit F23 described above acquires the solution MS and the algorithm AR using the graph data GD stored in the database DB.

[0285] The first acquisition unit F23 may have a function of determining a search range in the graph data GD. For example, the first acquisition unit F23 may determine the range surrounded by the dashed dotted line shown in FIG. 20 as the search range for the solution MS or the algorithm AR.

[0286] Here, a specific example of the acquisition of the solution MS and the algorithm AR by the first acquisition unit F23 will be described.

[0287] The first acquisition unit F23 acquires the solution MS and the algorithm AR using a natural language processing technique based on a large-scale language model. At this time, the large-scale language model has a predetermined context length that it can handle.

[0288] For example, the search range shown in FIG. 20 is excerpted and shown in FIG. 21, and the nodes ND are shown as parent factors A1 to A2, child factors B1 to B6, and grandchild factors C1 to C2 so that they can be distinguished from one another.

[0289] 21, the context lengths of the documents linked to each element are summed up and shown in parentheses as the "context length of the element." For example, if documents B1-1 and B1-2 are linked to child factor B1, and the context length of document B1-1 is 10k (=10×1000) and the context length of document B1-2 is 20k, the context length of child factor B1 is 30k (=10k+20k).

[0290] The first acquisition unit F23 determines a prompt to be input to the large-scale language model to obtain the solution MS and the algorithm AR based on the context length of each element.

[0291] For example, if the context length that can be handled by a large-scale language model is 100k, prompts are created by grouping factors so that the context length of each element is 100k or less.

[0292] 21, a first group G1 includes child factor B1 (30k), child factor B2 (15k), child factor B3 (20k), grandchild factor C1 (15k), and grandchild factor C2 (20k). The total context length in the first group G1 is 100k (=30k + 15k + 20k + 15k + 20k).

[0293] Furthermore, child factor B4 (10k), child factor B5 (40k), and child factor B6 (50k) are grouped into a second group G2. The total context length in the second group G2 is 100k (=10k+40k+50k).

[0294] The first group G1 and the second group G2 are indicated by two-dot chain lines in FIG.

[0295] The prompt for obtaining the solution MS with the first group G1 as the search range is defined as the first prompt PP1.

[0296] An example of the first prompt PP1 is shown in Figure 22. Note that the "context linked to child factor B1" in Figure 22 refers to image information, text information, etc., about the material linked to child factor B1. Furthermore, "text information" can include not only sentence information but also information expressed in the form of tables, graphs, etc.

[0297] The prompt for obtaining the solution MS with the second group G2 as the search range is defined as second prompt PP2. An example of the second prompt PP2 is shown in FIG.

[0298] As can be seen from Figures 21, 22 and 23, by limiting the context length of the materials referenced to obtain the solution MS and algorithm AR to a predetermined length (100k in this example), the solution MS can be appropriately acquired by the first acquisition unit F23.

[0299] The solutions MS and algorithms AR obtained for each group, such as the first group G1 and the second group G2, are temporarily stored in a memory or the like as appropriate.

[0300] In addition, after obtaining only the solution MS for all groups created according to the context length, the algorithm acquisition unit F23b of the first acquisition unit F23 may acquire an algorithm AR for each solution MS, and the prediction model acquisition unit F24a may acquire a prediction model PM for each solution MS.

[0301] The information on the solution MS, algorithm AR, and prediction model PM for each group obtained in this way is presented to the user by the information presentation unit F21a.

[0302] 24 and 25 show an example of processing executed by the CPU 11 of the server device 2 in this embodiment. The connection between the processing in both figures is indicated by a connector C4. Note that processing similar to that shown in FIG. 8 is given the same step number and description thereof will be omitted as appropriate.

[0303] The CPU 11 of the server device 2 accepts an input sentence input by the user using the user terminal 3 in step S101.

[0304] In step S102, the CPU 11 of the server device 2 analyzes the input sentence to understand the content of the sentence. This analysis is performed using, for example, natural language processing technology.

[0305] In step S103, the CPU 11 of the server device 2 determines whether the content of the input sentence presents a task SJ.

[0306] If it is determined that the content of the input sentence is not a request for problem solving (step S103: No), the CPU 11 of the server device 2A proceeds to step S104 and performs a corresponding process. The corresponding process differs depending on the analyzed content of the input sentence.

[0307] If it is determined that the content of the input statement is a request for problem solving (step S103: Yes determination), the CPU 11 of the server device 2 proceeds to step S201 and acquires a causal relationship diagram (e.g., a fishbone diagram) related to the problem SJ from the database DB. In this acquisition process, the CPU 11 of the server device 2A searches for similar problems SJ from past information and acquires the fishbone diagram used to solve the corresponding problem SJ.

[0308] In step S202, the CPU 11 of the server device 2 presents a fishbone diagram as a causal relationship diagram to the user.

[0309] In step S203, the CPU 11 of the server device 2 receives a user's evaluation of the presented fishbone diagram. This process can be said to be a process of receiving information on the user's evaluation indicating whether the fishbone diagram is appropriate for solving the problem SJ.

[0310] In step S108, the CPU 11 of the server device 2 determines whether the user evaluation is a positive evaluation. If the user evaluation is a negative evaluation (step S204: No), the CPU 11 of the server device 2 returns to step S201 and retrieves an unpresented fishbone diagram from the database DB.

[0311] On the other hand, if it is determined that the user evaluation is a positive evaluation (step S204: Yes judgment), the CPU 11 of the server device 2 proceeds to step S205 and generates graph data GD for the fishbone diagram that has received a positive evaluation from the user.

[0312] The graph data GD generation process converts the fishbone diagram to text to obtain text information. The graph data GD is then generated based on the text information. This process also involves linking materials stored in the database DB to each node ND in the graph data GD.

[0313] Next, the CPU 11 of the server device 2 proceeds to step S206 in Fig. 25 to determine the search range. In this process, as described above, the search range is determined according to the context length of each material.

[0314] In step S207, the CPU 11 of the server device 2 creates a prompt to be input to the AI ​​agent based on the search range.

[0315] In step S208, the CPU 11 of the server device 2 inputs the created prompt to the AI ​​agent to obtain answer information.

[0316] In step S209, the CPU 11 of the server device 2 performs a process of storing the acquired answer information in the RAM 13, the storage unit 19, or the like.

[0317] In step S210, the CPU 11 of the server device 2 determines whether or not a comprehensive search for answer information has been performed using the graph data GD. If there are any nodes ND or materials that have not been used to generate an answer among the nodes ND and materials that make up the graph data GD, the CPU 11 of the server device 2 determines "No" in step S210.

[0318] If it is determined that the comprehensive search has not been completed (step S210: No), the CPU 11 of the server device 2 proceeds to step S206 and sets the search ranges so as not to overlap, thereby generating different prompts and obtaining different answer information.

[0319] On the other hand, if it is determined that the comprehensive search has been completed (step S210: Yes judgment), the CPU 11 of the server device 2 proceeds to step S211 and generates a single aggregated answer information using one or more answer information stored in the RAM 13 or the memory unit 19.

[0320] In step S212, the CPU 11 of the server device 2 performs a process of presenting the collected answer information to the user.

[0321] After the process of step S212 shown in FIG. 25, the CPU 11 of the server device 2 may execute further processes.

[0322] For example, the CPU 11 of the server device 2 performs a process of obtaining evaluation information on whether the presented answer information is appropriate or not.

[0323] If a positive evaluation is obtained, the CPU 11 of the server device 2 proceeds to the process of step S105 in Fig. 8 and executes the process of acquiring the solution MS and the algorithm AR. That is, after completing the processes of Fig. 24 and Fig. 25, the CPU 11 of the server device 2 may further execute the series of processes from step S105 in Fig. 8 to step S123 in Fig. 10 if the answer information presented to the user is valid.

[0324] <10. Modifications> Several modifications will now be described.

[0325] The first modification is an example in which, in step S101 of FIG. 24, if an image is input as a causal relationship diagram when an input sentence is accepted, the process branches to execute each process from step S205 onwards.

[0326] Specifically, an image of the causal relationship diagram created by the user is received as input, and a series of processes are executed starting from step S205 in Fig. 24. An example of the flow of processes according to this modified example is shown in Fig. 26.

[0327] In step S102, the CPU 11 of the server device 2 analyzes the user input received in step S101.

[0328] The CPU 11 of the server device 2 performs branching processing in step S221 and step S103 depending on the analysis result.

[0329] Specifically, in step S221, the CPU 11 of the server device 2 determines whether the information input by the user is a causal relationship diagram such as a fishbone diagram. If it is determined that the information is not a causal relationship diagram (step S221: No), the CPU 11 of the server device 2 proceeds to step S103.

[0330] On the other hand, if it is determined that the inputted causal relationship diagram is a causal relationship diagram (step S221: Yes determination), the CPU 11 of the server device 2 proceeds to step S205 and generates graph data GD using the inputted causal relationship diagram.

[0331] It is preferable that the CPU 11 of the server device 2 stores the causal relationship diagram input by the user in the database DB before or after the process of step S205. This enables the CPU 11 of the server device 2 to obtain and present the newly added causal relationship diagram the next time the processes of steps S201 and S202 are performed.

[0332] Note that the user may input a causal relationship diagram as needed, at the prompt of the CPU 11 of the server device 2. For example, when the user determines that the causal relationship diagram presented by the CPU 11 of the server device 2 is inappropriate and there are no more causal relationship diagrams available to present to the user, the CPU 11 of the server device 2 may perform processing to prompt the user to create and input a causal relationship diagram.

[0333] The second modification is an example in which the user designates the search range instead of the CPU 11 of the server device 2 executing the process of determining the search range in step S206 of FIG.

[0334] An example of the flow of processing according to this modification is shown in Fig. 27. The series of processing shown in Fig. 27 is processing executed following each processing in Fig. 24 or Fig. 26. Of each processing, processing that differs from Fig. 25 will be mainly described.

[0335] After performing the process of generating the graph data GD in step S205, the CPU 11 of the server device 2 proceeds to the process of step S241.

[0336] In step S241, the CPU 11 of the server device 2 performs processing to present the generated graph data GD to the user.

[0337] Next, in step S242, the CPU 11 of the server device 2 accepts a user operation to specify a search range. This process is, for example, a process of identifying and acquiring child factors and grandchild factors selected by the user.

[0338] In step S243, the CPU 11 of the server device 2 determines whether the total of the context lengths of the materials included in the search range designated by the user falls within a predetermined range.

[0339] If it is determined that the total context length is not within the specified range and there is a problem (step S243: No determination), the CPU 11 of the server device 2 returns to the processing of step S242 and accepts the user's operation to specify the search range again.

[0340] On the other hand, if it is determined that the total context length is within the predetermined range and there is no problem (step S243: Yes determination), the CPU 11 of the server device 2 proceeds to the process of step S207.

[0341] The CPU 11 of the server device 2 creates a prompt based on the search range specified in step S207.

[0342] Furthermore, in step S208, the CPU 11 of the server device 2 inputs the created prompt to the AI ​​agent to obtain answer information.

[0343] In step S209, the CPU 11 of the server device 2 performs a process of storing the acquired answer information in the RAM 13, the storage unit 19, or the like.

[0344] In step S210, the CPU 11 of the server device 2 determines whether or not a comprehensive search for answer information using the graph data GD has been performed.

[0345] Note that instead of step S210, the user may be prompted to select whether or not to specify a different search range. If the user indicates an intention to specify a different search range, the CPU 11 of the server device 2 returns to step S242 again, accepts the operation to specify a new search range, and executes a series of processes subsequent to step S243.

[0346] This allows the user to obtain answer information by using the minimum range considered necessary and reducing the number of searches. Also, since the amount of answer information obtained by the user is reduced, the user can save time on examining and considering the answer information.

[0347] The user may specify the search range by a rough instruction. For example, if the user inputs a sentence indicating that a "detailed" examination is desired, the CPU 11 of the server device 2 sets the search range to include not only child factors and grandchild factors in the graph data GD but also great-grandchild factors and factors at deeper levels, in other words, factors that are farther away.

[0348] On the other hand, if the user inputs a sentence indicating that they desire a "simple" or "concise" examination, the CPU 11 of the server device 2 sets the search range to include only child factors or grandchild factors in the graph data GD, in other words, excluding factors that are far away.

[0349] Next, some examples of fishbone diagrams other than those mentioned above will be given. Figure 28 is an example of a fishbone diagram for a fault in a pipe through which a fluid flows.

[0350] Four parent factors are placed on the spine connected to the task SJ. Specifically, the fishbone diagram includes four parent factors: "Materials," "Design," "Assembly," and "Operation."

[0351] The fishbone diagram shown in FIG. 28 includes great-grandchild factors, which are lower-level concepts of grandchild factors.

[0352] Specifically, the child factors of the parent factor "design" are "stress concentration" and "state transition estimation."

[0353] The grandchild factors of the child factor "stress concentration" are "manufacturing variations in piping" and "estimated values ​​at the time of design."

[0354] Furthermore, the great-grandchild factors of the grandchild factor "estimated value at the time of design" are "1D CAE" and "3D CAE".

[0355] Here, "CAE" stands for "Computer Aided Engineering." "1D CAE" refers to one-dimensional models and methods that can evaluate the design from upstream to downstream, and is primarily used for functional design. "3D CAE" refers to three-dimensional models and methods that are used for more detailed analysis after the design has progressed to a certain extent, and reproduces realistic behavior.

[0356] In this way, a causal relationship diagram such as a fishbone diagram may include not only parent factors, child factors, and grandchild factors, but also great-grandchild factors and factors below them as a hierarchy of each factor.

[0357] FIG. 29 is an example of a fishbone diagram for an error on an app (application).

[0358] Three parent factors are arranged for the spine connected to the task SJ. Specifically, the fishbone diagram includes three parent factors: "Environment," "Design," and "Operation." In this way, the causal relationship diagram may be a fishbone diagram that includes three or fewer parent factors.

[0359] Note that a causal relationship diagram such as a fishbone diagram may be created using a large-scale language model. That is, when the user evaluates a fishbone diagram presented to the user negatively, the CPU 11 of the server device 2 may execute a process of creating a new fishbone diagram in the large-scale language model or a process of improving the fishbone diagram presented to the user into the large-scale language model.

[0360] 11. Application Example 4 The fourth application example is an example in which the database DB used in the first application example described above is constructed as a graph database.

[0361] The user inputs the following sentence as assignment SJ: "For a new image sensor equipped with pixels that adopt design type AA, please tell me a method to improve the saturation signal amount by 20% or more while relaxing the design margin of the image dark signal characteristics."

[0362] The CPU 11 of the server device 2 that realizes the function of constructing the database DB receives input of the task SJ and obtains, for example, a fishbone diagram as a causal relationship diagram.

[0363] Furthermore, if a positive evaluation is obtained from the user, the CPU 11 of the server device 2 generates graph data GD based on the fishbone diagram.

[0364] Furthermore, the CPU 11 of the server device 2 generates a prompt based on the search range that has been appropriately set in the graph data GD. If multiple search ranges are set, the CPU 11 generates as many prompts as there are search ranges.

[0365] The generated prompt includes, for example, the above-mentioned information D1, information D2, information D3, etc. Note that information D1, D2, and D3 correspond to material A and material B shown in FIG.

[0366] The CPU 11 of the server device 2 inputs the generated prompt to the AI ​​agent to obtain answer information.

[0367] The answer information obtained here includes information on the solution MS linked to information D1, which is the information described above regarding "optimization of the size of the photomask, which affects the implantation area in the ion implantation process," information on the solution MS linked to information D2, which is the information described above regarding "optimization of the impurity concentration in the ion implantation process to improve the saturation signal amount by 20%, and information on the solution MS linked to information D3, which is the information described above regarding "optimization of the impurity concentration in the ion implantation process to improve the saturation signal amount by 20%."

[0368] In obtaining these three solutions MS, one solution MS may be obtained by inputting one prompt, or three solution MS may be obtained by inputting one prompt.

[0369] Furthermore, the answer information obtained by inputting the prompt includes not only information on the solution method MS but also information on the algorithm AR as appropriate.

[0370] By inputting all the generated prompts, one or more solutions MS and algorithms AR are obtained as answer information. The CPU 11 of the server device 2 generates a single answer information that aggregates all the information based on these solutions MS and algorithms AR.

[0371] Replacing this one piece of answer information with the example mentioned above, it is information such as "In the past information for design type AA, the saturated signal amount is improved by optimizing the photomask. In the past information for design type AB and design type AC, the saturated signal amount is improved by optimizing the impurity concentration in the ion implantation process. In the past information for design type AA and design type AB, optimization was performed using reinforcement learning. In the past information for design type AC, optimization was performed using single-objective optimization."

[0372] The CPU 11 of the server device 2 presents this aggregated response information to the user along with information such as, "Based on the above information, we will use reinforcement learning to optimize the impurity concentration in the ion implantation process and present the results of our study. Is it okay to proceed?"

[0373] In this way, when a graph database is constructed as a database DB, the database DB can be used as the database DB used in the above-mentioned application examples 1, 2, and 3.

[0374] 12. Summary As described in the examples above, the information processing system 1 (1A, 1B, 1C) includes a reception unit F22 that receives input of a problem SJ from a user, a first acquisition unit F23 that acquires, from a database DB, a solution MS for solving the problem SJ received by the reception unit F22 and an algorithm AR for implementing the solution MS, a second acquisition unit F24 that acquires, from the database DB, a prediction model PM (PM1, PM2) suitable for the solution MS acquired by the first acquisition unit F23, and a solution acquisition processing unit F25 (F25B) that acquires a solution AS for the problem SJ using the algorithm AR to which execution conditions for the algorithm AR acquired by the first acquisition unit F23 are applied and the prediction model PM acquired by the second acquisition unit F24. The reception unit F22, for example, functions as an AI agent to interact with the user. Furthermore, the “solution MS” acquired by the first acquisition unit F23 is a method for solving the problem SJ. The "algorithm AR" acquired by the first acquisition unit F23 is a method for realizing the solution MS. The solution AS obtained based on the response to the large-scale language model may not be a physically or mathematically guaranteed solution AS. However, with this configuration, even when a large-scale language model is used, the solution MS and algorithm AR are selected from the database DB, and the algorithm AR is executed to obtain the solution AS. Therefore, since the database DB associates and stores the problem SJ, the solution MS, the algorithm AR, and the prediction model PM based on past performance, the presented algorithm AR is based on performance and is physically and mathematically guaranteed. Furthermore, the solution AS obtained by executing the algorithm AR can also be considered to be physically or mathematically guaranteed. Furthermore, the user can achieve their goal by directly applying the solution AS obtained using the information processing system to a real environment.

[0375] As described with reference to FIG. 9 and other figures, in the information processing system 1 (1A, 1B, 1C), if the solution MS acquired by the first acquisition unit F23 does not use the prediction model PM (PM1, PM2), the second acquisition unit F24 may not acquire the prediction model PM, and the solution acquisition processing unit F25 (F25B) may obtain the solution AS using an algorithm AR that applies execution conditions without using the prediction model PM. Some solutions MS do not require the prediction model PM. In such cases, the processing load on the information processing system 1 can be reduced by the second acquisition unit F24 not acquiring the prediction model PM. Furthermore, the solution acquisition processing unit F25 may execute an algorithm AR that does not require the use of the prediction model PM, thereby reducing the processing load on the information processing system 1 and obtaining an appropriate solution AS.

[0376] As described with reference to Fig. 4 etc., in the information processing system 1 (1A, 1B, 1C), the solution MS and the algorithm AR may be acquired from a database DB in which past tasks SJ and response results for the past tasks SJ are linked and stored. By extracting the solution MS and the algorithm AR based on the past response results, the obtained solution AS is physically or mathematically guaranteed.

[0377] As described with reference to FIG. 4 etc., in the information processing system 1 (1A, 1B, 1C), the prediction model PM (PM1, PM2) may be acquired from a database DB in which past tasks SJ and response results for the past tasks SJ are linked and stored. By extracting the prediction model PM based on the past response results, it becomes possible to obtain a physically or mathematically guaranteed solution AS.

[0378] As described with reference to FIG. 4 and other figures, in the information processing system 1 (1A, 1B, 1C), the prediction model PM (PM1, PM2) may include at least one of a physical prediction model, a chemical reaction model, and an inference model based on experimental data. This ensures that the data output from the prediction model PM is physically or mathematically guaranteed. Therefore, the solution AS obtained by the solution acquisition processing unit F25 (F25B) can also be physically or mathematically guaranteed.

[0379] 4 and the like, in the information processing system 1 (1A, 1B, 1C), the algorithm AR may include either an optimization algorithm for single-objective optimization or multi-objective optimization, or a reinforcement learning algorithm. By selecting an algorithm AR suitable for solving the problem SJ from among such multiple algorithms AR, it becomes possible to obtain a solution AS while reducing the processing load of the information processing system 1.

[0380] As described with reference to FIG. 3 and other figures, the information processing system 1 (1A, 1B, 1C) may include a summarization unit F26 that uses artificial intelligence or a platform model to summarize the results acquired by the first acquisition unit F23 and the second acquisition unit F24. For example, by summarizing the acquired information using a large-scale language model, the summarization unit F26 can present information in a format that is easy for the user to understand. For example, a large-scale language model may function as both the reception unit F22 and the summarization unit F26, and the large-scale language model may be configured to respond in a chat format, thereby providing information to the user through natural dialogue in a chat format. This allows the user to obtain information without performing any special operations, improving convenience.

[0381] As described with reference to Figures 3 and 4, the information processing system 1 (1A, 1B, 1C) includes a presentation processing unit F21 that performs a process of presenting information acquired by a first acquisition unit F23 to a user (step S106) and a process of presenting information acquired by a second acquisition unit F24 to a user (step S114). The reception unit F22 may receive a user's evaluation of the information presented by the presentation processing unit F21 (steps S107 and S115) and may execute at least one of a process of selecting a portion of the information acquired by the first acquisition unit F23 or a process of selecting a portion of the information acquired by the second acquisition unit F24 using the user's evaluation. The user's evaluation may be made, for example, by having the user select an optimal option from multiple options presented by the information processing system 1, or by having the user select whether or not to allow the use of the solution MS, algorithm AR, etc. selected by the information processing system 1. By accepting user evaluations of the solution methods MS, algorithms AR, or prediction models PM (PM1, PM2) presented by the information processing system 1, the information processing system 1 itself does not need to perform the process of selecting the optimal option from multiple options. For example, compared to a case in which the information processing system 1 comprehensively selects a huge number of combinations consisting of multiple solution methods MS, multiple algorithms AR, and multiple prediction models PM, obtains a solution AS by executing the algorithm AR for each combination, and compares these solutions AS to ultimately present an appropriate combination or solution AS, the processing burden on the information processing system 1 can be significantly reduced.

[0382] As described with reference to FIG. 3 and other figures, the solution acquisition processing unit F25 (F25B) in the information processing system 1 (1A, 1B, 1C) may generate an executable file for the algorithm AR and execute the executable file to obtain a solution AS. The executable file for the algorithm AR is written so that the algorithm AR is executed when execution conditions are satisfied. Therefore, by executing the executable file, the solution acquisition processing unit F25 can, for example, use the acquired solution MS, algorithm AR, and prediction model PM (PM1, PM2) and obtain an optimal solution from among multiple solutions AS obtained by setting parameters and the like within appropriate ranges.

[0383] 3 and the like, the information processing system 1 (1A, 1B, 1C) includes a presentation processing unit F21 that presents an executable file or execution conditions for the executable file to a user, and a reception unit F22 that receives a user's evaluation of the executable file or execution conditions presented by the presentation processing unit F21 and may cause a solution acquisition processing unit F25 (F25B) to generate an executable file using the user's evaluation. This prevents a solution AS obtained using conditions unintended by the user from being presented as an optimal solution.

[0384] As described with reference to FIGS. 11 and 12 , in the information processing system 1B, the solution acquisition processing unit F25B uses a software tool (algorithm execution tool) to obtain the solution AS. The software tool may be a tool that executes the algorithm AR using the prediction model PM (PM1, PM2). For example, the software tool may be a cloud application provided by another server device. This reduces the processing load on the information processing device (server device 2). Furthermore, the information processing device can obtain the solution AS simply by generating input data for the software tool, eliminating the need to generate an executable file for the algorithm AR.

[0385] As described with reference to FIG. 3 and the like, the information processing system 1 (1A, 1B, 1C) may include a presentation processing unit F21 that performs presentation processing on the solution AS obtained by the solution acquisition processing unit F25 (F25B). This allows the user to reflect the presented solution AS in improvements to the actual process, etc. Furthermore, when a large-scale language model performs the function of the presentation processing unit F21, the solution AS can be presented to the user through natural dialogue in a chat format. This allows the user to obtain the solution AS in a natural way through dialogue, thereby improving convenience.

[0386] As described with reference to Figures 13 and 14, the information processing system 1C includes multiple processors including a first processor, and the processing by the reception unit F22 may be executed by the first processor (e.g., CPU 11 of the server device 2C1), and the processing by the first acquisition unit F23 may be executed by a processor other than the first processor (e.g., CPU 11 of the server device 2C2). That is, the information processing system 1 may include multiple information processing devices (server devices 2C1, 2C2). The processing by the reception unit F22 and the processing by the first acquisition unit F23 may be executed by different information processing devices. This eliminates the need to excessively increase the performance of each information processing device, thereby enabling cost reductions.

[0387] As described with reference to Figures 13 and 14, the information processing system 1C includes multiple processors including a first processor, and the processing by the reception unit F22 may be executed by the first processor (e.g., CPU 11 of the server device 2C1), while the processing by the second acquisition unit F24 may be executed by a processor other than the first processor (e.g., CPU 11 of the server device 2C2). That is, the information processing system 1 may include multiple information processing devices (server devices 2C1, 2C2). The processing by the reception unit F22 and the processing by the second acquisition unit F24 may be executed by different information processing devices. This eliminates the need to excessively increase the performance of each information processing device, thereby enabling cost reduction.

[0388] As described with reference to Figures 13 and 14, the information processing system 1C includes multiple processors including a first processor, and the processing by the reception unit F22 may be executed by the first processor (e.g., CPU 11 of server device 2C1), while the processing by the solution acquisition processing unit F25 (F25B) may be executed by a processor other than the first processor (e.g., CPU 11 of server device 2C2). That is, the information processing system 1 may include multiple information processing devices (server devices 2C1, 2C2). The processing by the reception unit F22 and the processing by the solution acquisition processing unit F25 may be executed by different information processing devices. This eliminates the need to excessively increase the performance of each information processing device, thereby enabling cost reduction.

[0389] The information processing method of the present technology is one in which a processing device executes the following processes: a process of accepting input of a problem SJ by a user; a process of acquiring a solution MS for solving the accepted problem SJ and an algorithm AR for realizing the solution MS from a database DB; a process of acquiring a prediction model PM (PM1, PM2) suitable for the acquired solution MS from the database DB; and a process of obtaining a solution AS for the problem SJ using the algorithm AR to which execution conditions for the acquired algorithm AR are applied and the acquired prediction model PM.

[0390] In addition, the storage medium of the present technology stores a program that includes the following steps: a process for accepting input of a problem SJ by a user; a process for acquiring a solution MS for solving the accepted problem SJ and an algorithm AR for realizing the solution MS from a database DB; a process for acquiring a prediction model PM (PM1, PM2) suitable for the acquired solution MS from the database DB; and a process for obtaining a solution AS for the problem SJ using the algorithm AR to which execution conditions for the acquired algorithm AR are applied and the acquired prediction model PM, and the program is readable by a computer device.

[0391] Such an information processing method or storage medium can also provide the various functions and effects described above.

[0392] As described in the database generation method, the information stored in the database (DB), including the solution method (MS) and the algorithm (AR), may be updated through interaction with the user. For example, the information processing system 1 (1A, 1B, 1C) may include a reception unit (F22) that receives input of a task (SJ) from the user, a causal relationship diagram acquisition unit (information management unit (F27)) that acquires a causal relationship diagram (e.g., a fishbone diagram) from the database (DB) to be used to solve the task (SJ) received by the reception unit (F22), a graph data generation unit (F27b) that generates graph data (GD) having a graph structure consisting of nodes (ND) and edges (ED) based on the causal relationship diagram, a prompt generation unit (first acquisition unit (F23)) that sets one or more search ranges for obtaining answer information based on the graph data (GD) and generates a prompt for each search range, a solution acquisition unit (F23a) that obtains a solution (MS) for each prompt, and a presentation processing unit (F21) that presents the solution (MS). The graph data (GD) stored in the database is configured by linking materials to each node (ND). Such graph data (GD) includes distance information. Furthermore, the graph data GD can be presented with a visual representation of distance information, making it easy for users to understand and suitable for use as a GUI. Multiple search ranges can be set. By setting all nodes ND in the graph data GD to be included in one of the search ranges along with the documents, it is possible to comprehensively examine each factor included in the causal relationship diagram, providing the user with useful information (solution methods MS and algorithms AR). Furthermore, documents such as case studies and solution proposals stored as independent information in the database DB are linked to parent and child factors during the graph data GD generation process and organized into systematic data. Furthermore, when a user inputs a task SJ, this graph data GD is reconstructed using new documents added since the previous generation of the graph data GD. This ensures that the graph data GD presented to the user in response to the user's input of the task SJ is always up to date.That is, after a new solution MS is developed to solve the problem SJ and the materials for that solution are stored in the database DB, the graph data GD can be reconstructed as appropriate, making it possible to present various information to the user, including the latest solution MS. Also, in the information processing system 1 (1A, 1B, 1C), by storing prompt inputs to the AI ​​agent and the obtained information as a cache, it is possible to present solutions MS and algorithms AR for similar problems SJ while reducing token consumption for a certain period of time, or until new materials are added to the database.

[0393] The solution acquisition unit F23a in the information processing system 1 (1A, 1B, 1C) may obtain the solution MS using artificial intelligence or a board model. By using artificial intelligence or a board model, it is possible to efficiently perform a process of linking factors as each node ND in the causal relationship diagram and materials individually stored in the database DB according to the degree of affinity. Then, it is possible to acquire and present an appropriate solution MS or algorithm AR according to the linking results.

[0394] In the information processing system 1 (1A, 1B, 1C), the search range may be set to include at least some of the nodes ND in the graph data GD, and the prompt generation unit (first acquisition unit F23) may set the search range based on the context length related to the nodes ND. In particular, when the information processing system 1 (1A, 1B, 1C) inputs a prompt to an artificial intelligence or a circuit board model to obtain a solution MS or an algorithm AR, the search range may be determined taking into account a context length that the artificial intelligence or the circuit board model can handle, thereby increasing the certainty of obtaining the solution MS or the algorithm AR.

[0395] In the information processing system 1 (1A, 1B, 1C), each of all nodes ND included in the graph data GD may be included in at least one search range, thereby allowing the information processing system 1 (1A, 1B, 1C) to obtain a comprehensive answer using all nodes ND included in the graph data GD.

[0396] In the information processing system 1 (1A, 1B, 1C), the reception unit F22 may receive input of a causal relationship diagram from a user, and the causal relationship diagram acquisition unit (information management unit F27) may acquire the causal relationship diagram input by the user instead of acquiring the causal relationship diagram from the database DB. This makes it possible to search for and present solutions MS and algorithms AR using causal relationship diagrams not stored in the database DB. Furthermore, by storing the causal relationship diagram provided by the user in the database DB, it can be selected as information to be presented in response to the user's input of a task SJ from the next time onwards. This increases the likelihood of presenting an appropriate causal relationship diagram in response to the user's input of a task SJ.

[0397] The effects described in this specification are merely examples and are not limiting, and other effects may also be present.

[0398] Furthermore, the above-described examples may be combined in any manner, and even when various combinations are used, the various effects described above can be obtained.

[0399] <13. The Present Technology> The present technology may also be configured as follows: (1) An information processing system comprising: a reception unit that receives a problem input by a user; a first acquisition unit that acquires, from a database, a solution for solving the problem received by the reception unit and an algorithm for realizing the solution; a second acquisition unit that acquires, from the database, a predictive model suitable for the solution acquired by the first acquisition unit; and a solution acquisition processing unit that acquires a solution to the problem using the algorithm to which execution conditions for the algorithm acquired by the first acquisition unit are applied and the predictive model acquired by the second acquisition unit. (2) The information processing system described in (1) above, in which, when the solution acquired by the first acquisition unit does not use the predictive model, the second acquisition unit does not acquire the predictive model, and the solution acquisition processing unit acquires the solution using the algorithm to which the execution conditions are applied without using the predictive model. (3) The information processing system described in any of (1) to (2) above, in which the solution and the algorithm are acquired from a database in which past problems and results of addressing the past problems are stored in association with each other. (4) The information processing system according to any one of (1) to (3), wherein the prediction model is acquired from a database in which past problems and response results to the past problems are linked and stored. (5) The information processing system according to any one of (1) to (4), wherein the prediction model includes at least one of a physical prediction model, a chemical reaction model, and an inference model based on experimental data. (6) The information processing system according to any one of (1) to (5), wherein the algorithm includes one of an optimization algorithm that performs single-objective optimization or multi-objective optimization, and a reinforcement learning algorithm. (7) The information processing system according to any one of (1) to (6), further comprising a summarization unit that summarizes the results acquired by the first acquisition unit and the results acquired by the second acquisition unit using artificial intelligence or a base model.(8) The information processing system according to any of (1) to (7) above, further comprising: a presentation processing unit that performs a process of presenting information acquired by the first acquisition unit to a user; and a process of presenting information acquired by the second acquisition unit to a user, wherein the reception unit receives a user's evaluation of the information presented by the presentation processing unit, and causes at least one of a process of selecting a portion of the information acquired by the first acquisition unit and a process of selecting a portion of the information acquired by the second acquisition unit using the user's evaluation. (9) The information processing system according to any of (1) to (8) above, further comprising: a presentation processing unit that presents the executable file or the execution conditions for the executable file to a user, wherein the reception unit receives a user's evaluation of the executable file or the execution conditions presented by the presentation processing unit, and causes the solution acquisition processing unit to generate the executable file using the user's evaluation. (11) The information processing system according to any of (1) to (10) above, wherein the solution acquisition processing unit uses a software tool to obtain the solution, and the software tool is a tool that executes the algorithm using the predictive model. (12) The information processing system according to any of (1) to (11) above, comprising a presentation processing unit that performs presentation processing on the solution obtained by the solution acquisition processing unit. (13) The information processing system according to any of (1) to (12) above, comprising a plurality of processors including a first processor, wherein processing by the reception unit is executed by the first processor, and processing by the first acquisition unit is executed by a processor other than the first processor. (14) The information processing system according to any of (1) to (13) above, comprising a plurality of processors including a first processor, wherein processing by the reception unit is executed by the first processor, and processing by the second acquisition unit is executed by a processor other than the first processor.(15) An information processing system according to any one of (1) to (14) above, comprising a plurality of processors including a first processor, wherein processing by the reception unit is executed by the first processor, and processing by the solution acquisition processing unit is executed by a processor other than the first processor. (16) An information processing system comprising: a reception unit that receives input of a problem from a user; a causal relationship diagram acquisition unit that acquires from a database a causal relationship diagram to be used for solving the problem received by the reception unit; a graph data generation unit that generates graph data having a graph structure consisting of nodes and edges based on the causal relationship diagram; a prompt generation unit that sets one or more search ranges for obtaining answer information based on the graph data and generates a prompt for each of the search ranges; a solution acquisition unit that obtains a solution for each of the prompts; and a presentation processing unit that presents the solution. (17) The information processing system according to (16), wherein the solution acquisition unit obtains the solution using artificial intelligence or a foundation model. (18) The information processing system according to any one of (16) to (17), wherein the search range is set to include at least some of the nodes in the graph data, and the prompt generation unit sets the search range based on a context length related to the node. (19) The information processing system according to (18), wherein all of the nodes included in the graph data are each included in at least one of the search ranges. (20) The information processing system according to any one of (16) to (19), wherein the reception unit receives input of the causal relationship diagram by a user, and the causal relationship diagram acquisition unit acquires the causal relationship diagram input by the user instead of acquiring the causal relationship diagram from the database.

[0400] 1, 1A, 1B, 1C Information processing system 11 CPU AR Algorithm AS Solution DB Database F21 Presentation processing unit F22 Reception unit F23 First acquisition unit F24 Second acquisition unit F25, F25B Solution acquisition processing unit F26 Summarization unit MS Solution PM, PM1, PM2 Prediction model SJ Task

Claims

1. An information processing system comprising: a reception unit that receives input of a problem from a user; a first acquisition unit that acquires from a database a solution to the problem received by the reception unit and an algorithm for realizing the solution; a second acquisition unit that acquires from the database a predictive model suitable for the solution acquired by the first acquisition unit; and a solution acquisition processing unit that obtains a solution to the problem using the algorithm to which execution conditions for the algorithm acquired by the first acquisition unit are applied and the predictive model acquired by the second acquisition unit.

2. The information processing system of claim 1, wherein, when the solution acquired by the first acquisition unit does not use the predictive model, the second acquisition unit does not acquire the predictive model, and the solution acquisition processing unit obtains the solution using the algorithm to which the execution conditions are applied without using the predictive model.

3. The information processing system according to claim 1, wherein the solution and the algorithm are obtained from a database in which past problems and the results of responses to the past problems are linked and stored.

4. The information processing system according to claim 1, wherein the predictive model is obtained from a database in which past issues and the results of responses to the past issues are linked and stored.

5. The information processing system according to claim 1, wherein the prediction models include at least one of a physical prediction model, a chemical reaction model, and an inference model based on experimental data.

6. The information processing system according to claim 1, wherein the algorithm includes either an optimization algorithm for single-objective optimization or multi-objective optimization, or a reinforcement learning algorithm.

7. The information processing system according to claim 1, further comprising a summarizing unit that summarizes the results obtained by the first acquisition unit and the results obtained by the second acquisition unit using artificial intelligence or a base model.

8. An information processing system as described in claim 1, further comprising a presentation processing unit that performs a process of presenting information acquired by the first acquisition unit to a user and a process of presenting information acquired by the second acquisition unit to a user, wherein the reception unit receives a user's evaluation of the information presented by the presentation processing unit and uses the user's evaluation to perform at least one of a process of selecting a portion of the information acquired by the first acquisition unit and a process of selecting a portion of the information acquired by the second acquisition unit.

9. The information processing system according to claim 1, wherein the solution acquisition processing unit generates an executable file for the algorithm and executes the executable file to obtain the solution.

10. An information processing system as described in claim 9, further comprising a presentation processing unit that presents the executable file or the execution conditions for the executable file to a user, wherein the reception unit receives a user's evaluation of the executable file or the execution conditions presented by the presentation processing unit, and causes the solution acquisition processing unit to generate the executable file using the user's evaluation.

11. The information processing system according to claim 1, wherein the solution acquisition processing unit uses a software tool to obtain the solution, and the software tool is a tool that executes the algorithm using the prediction model.

12. The information processing system according to claim 1, further comprising a presentation processing unit that performs presentation processing on the solution obtained by the solution acquisition processing unit.

13. The information processing system according to claim 1, comprising a plurality of processors including a first processor, wherein processing by the reception unit is executed by the first processor, and processing by the first acquisition unit is executed by a processor other than the first processor.

14. The information processing system according to claim 1, comprising a plurality of processors including a first processor, wherein processing by the reception unit is executed by the first processor, and processing by the second acquisition unit is executed by a processor other than the first processor.

15. The information processing system according to claim 1, comprising a plurality of processors including a first processor, wherein processing by the reception unit is executed by the first processor, and processing by the solution acquisition processing unit is executed by a processor other than the first processor.

16. An information processing system comprising: a reception unit that receives input of a problem from a user; a causal relationship diagram acquisition unit that acquires from a database a causal relationship diagram to be used to solve the problem received by the reception unit; a graph data generation unit that generates graph data having a graph structure consisting of nodes and edges based on the causal relationship diagram; a prompt generation unit that sets one or more search ranges for obtaining answer information based on the graph data and generates a prompt for each search range; a solution acquisition unit that obtains a solution for each prompt; and a presentation processing unit that presents the solution.

17. The information processing system according to claim 16, wherein the solution acquisition unit acquires the solution using artificial intelligence or a base model.

18. An information processing system as described in claim 16, wherein the search range is set to include at least some of the nodes in the graph data, and the prompt generation unit sets the search range based on a context length related to the nodes.

19. The information processing system according to claim 18, wherein each of all the nodes included in the graph data is included in at least one of the search ranges.

20. The information processing system of claim 16, wherein the reception unit receives input of the causal relationship diagram by a user, and the causal relationship diagram acquisition unit acquires the causal relationship diagram input by the user instead of acquiring the causal relationship diagram from the database.

Images