Information processing system

Abstract

To improve the response accuracy of large-scale language models. [Solution] The information processing system (1) includes, when a first question sentence relating to one of the multiple processes included in a business flow is input, an extraction means (111) for extracting first text data related to the first question sentence from one workspace corresponding to one process in a database (30) which includes multiple workspaces (WS) corresponding to each of the multiple processes; an input means (112) for inputting the first question sentence and the first text data into a large-scale language model; and an acquisition means (113) for obtaining a first answer to the first question sentence generated by the large-scale language model.

Description

Technical Field

[0001] The present invention relates to the technical field of information processing systems.

Background Art

[0002] As this type of system, for example, a system that generates a document to be input as a prompt to a large language model (LLM) has been proposed (see Patent Document 1).

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] For example, a chatbot using a mechanism (Retrieval-Augmented Generation: RAG) that gives a large language model a unique information source by combining the large language model with the search of a specific information source (hereinafter, appropriately referred to as a "knowledge base") has been proposed. Such a chatbot may be used as a business support tool. By the way, in a company, a business process is often defined for a certain business. When the above chatbot is used as a business support tool, if no measures are taken, there is a high possibility that an answer ignoring the business process will be generated. That is, the answer accuracy of the large language model may decrease. Note that a large language model is a language model constructed using a very large dataset and deep learning technology.

[0005] The present invention has been made in view of the above problems, and an object thereof is to provide an information processing system that can improve the answer accuracy of a large language model. [Means for solving the problem]

[0006] An information processing system according to one aspect of the present invention includes, when a first question statement relating to one of a plurality of processes included in a business flow is input, an extraction means for extracting first text data related to the first question statement from a workspace corresponding to one of the plurality of processes in a database which includes a plurality of workspaces corresponding to each of the plurality of processes; an input means for inputting the first question statement and the first text data into a large-scale language model; and an acquisition means for obtaining a first answer to the first question statement generated by the large-scale language model. [Brief explanation of the drawing]

[0007] [Figure 1] This is a diagram showing the configuration of an information processing system according to an embodiment. [Figure 2] This block diagram shows an example of the configuration of a computing device according to the embodiment. [Figure 3] This figure shows an example of a displayed image. [Modes for carrying out the invention]

[0008] Embodiments of the information processing system will be described with reference to Figures 1 to 3. In Figure 1, the information processing system 1 comprises an information processing device 10, a server 20, and a database 30. The information processing device 10, the server 20, and the database 30 are configured to communicate with each other via a network NW.

[0009] Server 20 and database 30 provide a chatbot using RAG. Server 20 is a server for operating a Large-Scale Language Model (LLM). Therefore, Server 20 may be referred to as an LLM server. Server 20 may also be a cloud server.

[0010] Database 30 includes multiple workspaces WS. Database 30 may be implemented by a single device (e.g., a server) or by multiple devices. In this embodiment, each of the multiple workspaces WS corresponds to a process included in the business flow. One workspace in the multiple workspaces WS (e.g., workspace WS1) contains text data related to one of the processes.

[0011] For example, the text data contained in (in other words, registered in) database 30 may be fragmented data generated by splitting a document. Fragmented data may be called "chunks". Methods for splitting a document include, for example, splitting at a fixed length, splitting at the sentence level based on sentence delimiters, or splitting based on structure such as Markdown. Multiple fragmented data may be vectorized and registered in database 30. In other words, database 30 may be a vector database / vector store. Database 30 may also be called a knowledge base.

[0012] The information processing device 10 comprises an arithmetic unit 11, a storage device 12, a communication device 13, an input device 14, and an output device 15. The arithmetic unit 11, storage device 12, communication device 13, input device 14, and output device 15 are connected via a data bus 16. The information processing device 10 may be a personal computer, a tablet terminal, or a smartphone.

[0013] The arithmetic unit 11 may have a processor. The arithmetic unit 11 may have a single processor or multiple processors. In other words, the arithmetic unit 11 may have one or more processors. Furthermore, the processor may be a multi-core processor. If the arithmetic unit 11 has a single processor that is a multi-core processor, then logically, the arithmetic unit 11 can be said to have multiple processors.

[0014] The processor may be at least one of the following: CPU (Central Processing Unit), GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), and TPU (Tensor Processing Unit).

[0015] The storage device 12 may be at least one of the following: RAM (Random Access Memory), ROM (Read Only Memory), hard disk drive, magneto-optical disk drive, SSD (Solid State Drive), and optical disk array. In other words, the storage device 12 may be implemented by a single device or by multiple devices.

[0016] The communication device 13 may be capable of communicating with an external device (for example, a server 20) of the information processing device 10. The communication device 13 may use either wired or wireless communication.

[0017] The input device 14 is a device capable of receiving information input to the information processing device 10 from an external source. The input device 14 may include an operating device (e.g., a keyboard, mouse, touch panel, etc.) that can be operated by the user of the information processing device 10. The input device 14 may include a recording medium reader capable of reading information recorded on a recording medium that can be attached to and detached from the information processing device 10, such as a USB (Universal Serial Bus) memory. When information is input to the information processing device 10 via the communication device 13 (in other words, when the information processing device 10 acquires information via the communication device 13), the communication device 13 may function as an input device.

[0018] The output device 15 is a device capable of outputting information to the outside of the information processing device 10. The output device 15 has a display device 151 capable of outputting visual information such as characters and images as the above information. The output device 15 may also have a speaker capable of outputting auditory information such as sound as the above information. The output device 15 may also have a vibration motor capable of outputting tactile information such as vibration as the above information. The output device 15 may also have a printer. The output device 15 may be capable of outputting information to a recording medium that can be attached to and detached from the information processing device 10, such as a USB memory stick. When the information processing device 10 outputs information via the communication device 13, the communication device 13 may function as an output device.

[0019] The storage device 12 is capable of storing desired data. The storage device 12 may store the computer program CP that the arithmetic unit 11 will execute. The storage device 12 may temporarily store data that the arithmetic unit 11 will use temporarily when the arithmetic unit 11 is executing the computer program CP.

[0020] Furthermore, the computer program CP may be recorded on a non-temporary recording medium that is readable by a computer. In this case, the computer program CP may be stored in the storage device 12 by reading the recording medium using a recording medium reading device (not shown) provided by the information processing device 10. Furthermore, at least one of the following may be used as the recording medium: an optical disc, a magnetic medium, a magneto-optical disc, a semiconductor memory, and any other medium capable of storing a program. Furthermore, the computer program CP may be obtained from an external device (not shown) of the information processing device 10 via a communication device 13. In other words, the computer program CP may be downloaded from an external device to the storage device 12 of the information processing device 10.

[0021] The arithmetic unit 11 (e.g., a processor) may execute the processing that the information processing apparatus 10 should perform, together with the storage device 12 in which the computer program CP is stored (in other words, together with the storage device 12 and the computer program CP stored in the storage device 12). For example, by executing the computer program CP, a logical functional block for executing the processing that the information processing apparatus 10 should perform may be realized in the arithmetic unit 11 (e.g., within the processor).

[0022] The arithmetic unit 11 of the information processing apparatus 10 has a search unit 111, an input unit 112, an acquisition unit 113, and a generation unit 114 in order to use the chatbot provided by the server 20 and the database 30 (see FIG. 2). The search unit 111, the input unit 112, the acquisition unit 113, and the generation unit 114 may be realized as the above-described logical functional blocks. Note that at least one of the search unit 111, the input unit 112, the acquisition unit 113, and the generation unit 114 may be realized as a physical processing circuit. At least one of the search unit 111, the input unit 112, the acquisition unit 113, and the generation unit 114 may be realized in a form in which logical functional blocks and physical processing circuits are mixed.

[0023] For example, the user U may use the chatbot using the information processing apparatus 10. For example, the user U may input a question sentence regarding one process included in the business process to the chatbot using the information processing apparatus 10. In this case, the user U may input the first question sentence via the input device 14 of the information processing apparatus 10. At this time, the user U may input via the input device 14 that the first question sentence is a question sentence regarding the above one process. For example, an image showing a plurality of processes included in the business process may be displayed on the display device 151. The user U may input the process related to the first question sentence by selecting one process in the image.

[0024] The term "question sentence" is not limited to interrogative sentences. For example, a "question sentence" may include expressions such as requests, instructions, or commands, such as "Tell me about ****" or "Answer me about ****." Therefore, "question sentence" is a concept that includes not only sentences in the form of interrogative sentences, but also sentences that include expressions such as requests, instructions, or commands. In other words, a "question sentence" can mean a sentence that seeks an answer from the other party.

[0025] The search unit 111 of the information processing device 10 searches for a workspace WS1 corresponding to a single process among multiple workspaces WS contained in the database 30, based on the first query statement. For example, the search unit 111 may calculate a search score indicating the degree of association between the first query statement and the text data contained in workspace WS1. The search unit 111 may extract text data having a search score of a predetermined value or higher as text data related to the first query statement. The search unit 111 may extract multiple text data related to the first query statement. In other words, the search unit 111 may extract one or more text data related to the first query statement from workspace WS1. Various existing methods can be applied to the calculation method of the search score. Therefore, a detailed explanation of the calculation method of the search score is omitted.

[0026] The input unit 112 of the information processing device 10 sends a prompt to the server 20 via the communication device 13, which includes a first question and text data related to the first question. As a result, the prompt is input to the large-scale language model. The server 20 sends the first answer to the first question, generated by the large-scale language model, to the information processing device 10. The acquisition unit 113 of the information processing device 10 acquires the first answer sent by the server 20. The arithmetic unit 11 of the information processing device 10 may control the display device 151 to display the first answer.

[0027] After the acquisition unit 113 acquires the first answer, the generation unit 114 of the information processing device 10 generates a second question sentence regarding the next step in the first step, according to the first answer. In other words, the second question sentence is generated automatically. The generation unit 114 may generate the second question sentence according to the first answer and the first question sentence.

[0028] For example, the business flow may be a business flow related to legal certification. In this case, the business flow may include the steps of "searching for relevant legal texts" and "interpreting legal texts." For example, the first question above may be a question related to the "searching for relevant legal texts" step. In this case, the answer to the first question above may include the content of the legal text. The step following the "searching for relevant legal texts" step may be the "interpretation of legal texts" step. For example, the generation unit 114 may generate a question such as "What is ****?" as the second question above, depending on the content of the legal text included in the first answer, to ask for the interpretation of words included in the legal text.

[0029] After the second query is generated, the search unit 111 searches for the workspace WS2 corresponding to the next step in one of the multiple workspace WS contained in the database 30 based on the second query. For example, the search unit 111 may calculate a search score indicating the degree of association between the second query and the text data contained in the workspace WS2. The search unit 111 may extract text data having a search score of a predetermined value or higher as text data related to the second query.

[0030] The input unit 112 sends a prompt to the server 20 via the communication device 13, which includes a second question and text data related to the second question. As a result, the prompt is input to the large-scale language model. The server 20 sends the second answer to the second question, generated by the large-scale language model, to the information processing device 10. The acquisition unit 113 of the information processing device 10 acquires the second answer sent by the server 20. The arithmetic unit 11 may control the display device 151 to display the second answer.

[0031] For example, the display device 151 may display the image 200 shown in Figure 3. The image 200 may include an area 201 for displaying the first question, an area 202 for displaying the first answer, an area 203 for displaying the second question, and an area 204 for displaying the second answer.

[0032] After the second question and second answer are displayed on the display device 151, user U may modify the second question via the input device 14. In this case, the input means 14 may accept input from user U to modify the second question. If the second question is modified by user U, the search unit 111 searches the workspace WS2 based on the modified second question. For example, the search unit 111 may calculate a search score indicating the degree of association between the modified second question and the text data contained in the workspace WS2. The search unit 111 may extract text data with a search score of a predetermined value or higher as text data related to the modified second question.

[0033] The input unit 112 sends a prompt to the server 20 via the communication device 13, which includes the modified second question and text data related to the modified second question. As a result, the prompt is input to the large-scale language model. The server 20 sends the third answer to the modified second question, generated by the large-scale language model, to the information processing device 10. The acquisition unit 113 of the information processing device 10 acquires the third answer sent by the server 20. The arithmetic unit 11 may control the display device 151 to display the third answer.

[0034] (Technical effects) As described above, the database 30 includes multiple workspaces WS, each corresponding to one of the multiple processes included in the business flow. The search unit 111 of the information processing device 10 searches for a workspace WS1 corresponding to one of the processes related to the first question entered by user U. Workspace WS1 contains text data related to that process. Therefore, by searching workspace WS1, the search unit 111 extracts the first question and the text data related to that process. The input unit 112 of the information processing device 10 sends a prompt to the server 20 containing the first question and the text data related to the first question and that process (in other words, inputs the prompt to the large-scale language model). Therefore, the first answer generated by the large-scale language model can be expected to be an answer related to that process. Accordingly, the information processing system 1 according to this embodiment can generate an appropriate answer to a question specific to one of the processes included in the business flow. As a result, the information processing system 1 according to this embodiment can improve the answer accuracy of the large-scale language model.

[0035] Various aspects of the invention derived from the embodiments described above are described below.

[0036] An information processing system according to one aspect of the invention includes, when a first question sentence relating to one of a plurality of processes included in a business flow is input, an extraction means for extracting first text data related to the first question sentence from a workspace corresponding to one of the plurality of processes in a database which includes a plurality of workspaces corresponding to each of the plurality of processes; an input means for inputting the first question sentence and the first text data into a large-scale language model; and an acquisition means for acquiring a first answer to the first question sentence generated by the large-scale language model. In the above embodiment, the "search unit 111" corresponds to an example of the "extraction means", the "input unit 112" corresponds to an example of the "input means", and the "acquisition unit 113" corresponds to an example of the "acquisition means".

[0037] In the information processing system according to the above embodiment, the extraction means may extract second text data relating to at least the second question sentence corresponding to the first answer from another workspace of the database corresponding to the next step of the first step, the input means may input the second question sentence and the second text data into the large-scale language model, and the acquisition means may acquire the second answer to the second question sentence generated by the large-scale language model.

[0038] The information processing system according to the above embodiment may include a receiving means for receiving user input and a display means. After the second question and the second answer are displayed by the display means, if the receiving means receives user input to modify the second question, the extraction means may extract third text data related to the modified second question from the other workspace, and the input means may input the modified second question and the third text data into the large-scale language model. In the above embodiment, the "input device 14" corresponds to an example of the "receiving means," and the "display device 151" corresponds to an example of the "display means."

[0039] The present invention is not limited to the embodiments described above, and can be modified as appropriate without contradicting the gist or idea of ​​the invention as can be read from the claims and specification as a whole. Information processing systems involving such modifications are also included within the technical scope of the present invention. [Explanation of symbols]

[0040] 1... Information processing system, 10... Information processing device, 20... Server, 30... Database, 111... Search unit, 112... Input unit, 113... Acquisition unit, 114... Generation unit

Claims

1. When a first question statement relating to one of several processes included in a business workflow is entered, an extraction means extracts first text data related to the first question statement from one workspace corresponding to the one process in a database containing multiple workspaces corresponding to each of the multiple processes, An input means for inputting the first question sentence and the first text data into a large-scale language model, An acquisition means for obtaining a first answer to the first question sentence generated by the large-scale language model, An information processing system equipped with the following features.

2. The extraction means extracts second text data from another workspace in the database corresponding to the next step after the first step, relating to at least the second question sentence corresponding to the first answer. The input means inputs the second question sentence and the second text data into the large-scale language model. The acquisition means acquires the second answer to the second question sentence, which is generated by the large-scale language model. The information processing system according to claim 1.

3. A means of receiving user input, Display means and Equipped with, After the second question and the second answer are displayed by the display means, if the receiving means receives user input to modify the second question, the extraction means extracts third text data related to the modified second question from the other workspace. The input means inputs the modified second question sentence and the third text data into the large-scale language model. The information processing system according to claim 2.

Images