Information processing apparatus, control method of information processing apparatus, program, and information processing system

Abstract

To provide a mechanism capable of easily searching for information indicating a file storage location without being restricted by circumstances under which the file has been stored on a network.SOLUTION: A user terminal 101, which can be connected via a network 105 to a large language model server 103 in which a large language model searches for URLs being information indicating file storage locations on the basis of similarity of language vectors of natural language, comprises a CPU 201 that transmits, in natural language, a query sentence for inquiring about a file storage location to the large language model, and receives, from the large language model, a response sentence that conveys a URL being information indicating the file storage location searched by the large language model using a language vector of the natural language of the query sentence.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an information processing device, a control method for an information processing device, a program, and an information processing system. [Background technology]

[0002] One well-known method for natural language search is to use large-scale language models. One such method involves converting a large amount of natural language data into language vectors in advance, calculating the similarity between the vectors and a query sentence, also written in natural language, and selecting search results with high similarity. However, in the offices of knowledge workers and other users, users often face difficulties when searching for a specific file among a vast number of internal document files. However, if a user wants to search for a file on the Internet outside the company using a personal computer (PC), there are file search methods available, such as web search engines.

[0003] However, when a user searches for a file that they once viewed on their PC, they may find it difficult to remember whether the file was stored on a website on the company's intranet, on a file server, or on their own PC. Even if a user remembers that a file they once viewed on their PC is stored on a file server, they may find it difficult to remember where on the file server the file was stored. This is also true when a user uses an image forming device that transmits scanned files of internal company documents to a file server via SMB (Server Message Block) or other communication protocols. In other words, there are many situations in the office where it is difficult for users to find information indicating the storage location of a file.

[0004] As a related technique, Patent Document 1 discloses a technique in which a file is uploaded to a file server via a GUI of a file management system. In the technique disclosed in Patent Document 1, the file management system extracts text from the file name and file contents, and when a user asks a question, it outputs a file with the extracted text corresponding to the question as an icon on the GUI. [Prior art documents] [Patent documents]

[0005] [Patent Document 1] Japanese Patent Application Publication No. 2023-62251 Summary of the Invention [Problem to be solved by the invention]

[0006] However, the technology disclosed in Patent Document 1 only searches for specific files that have been uploaded via a file management system, which poses a problem in that the technology disclosed in Patent Document 1 cannot search for in-house files when a file is uploaded via a browser or when a file is stored on a file server via SMB from an image forming device rather than a PC.

[0007] The present invention has been made in view of the above-mentioned problems, and aims to provide a mechanism that enables easy retrieval of information indicating the storage location of a file, regardless of the circumstances under which the file was stored on a network. [Means for solving the problem]

[0008] In order to achieve the above-mentioned object, the information processing device of the present invention is an information processing device that can be connected via a network to a server on which a large-scale language model operates that searches for information indicating the storage location of a file based on the similarity of language vectors of natural language, and is characterized by comprising: a message sending means that sends a query message in natural language to the large-scale language model to inquire about the storage location of the file; and a message receiving means that receives from the large-scale language model a response message that conveys information indicating the storage location of the file that the large-scale language model has searched for using the language vector of the natural language of the query message. [Effects of the Invention]

[0009] According to the present invention, it is possible to easily search for information indicating the storage location of a file, without being restricted by the circumstances under which the file was stored on the network. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 1 is a diagram illustrating a configuration of an information processing system according to a first embodiment. [Figure 2] FIG. 2 is a block diagram showing an example of a hardware configuration common to a user terminal, a file server, and a large-scale language model server in the first and second embodiments. [Figure 3A] FIG. 2 is a diagram showing a database constructed in a large-scale language model server in the first embodiment. [Figure 3B] 10 is a flowchart showing the operation of the large-scale language model server in the first embodiment when a user terminal performs an operation on a file and the URL of the file is sent to the large-scale language model server by a resident program of the user terminal. [Figure 3C] 10 is a flowchart showing the operation of the large-scale language model server in the first embodiment when a question sentence and a user ID are sent to the large-scale language model server through a chat program by a resident program on a user terminal. [Figure 4A]10 is a flowchart showing the operation of a user terminal when the user terminal performs a file-related operation on the user terminal itself or the file server in the first embodiment. [Figure 4B] FIG. 10 is a diagram showing a screen displayed on the display unit of a user terminal when the user terminal sends a question about the storage location of a file to a large-scale language model server in the first embodiment. [Figure 4C] 10 is a flowchart showing the operation of a user terminal in the first embodiment when a question sentence and a user ID are sent to a large-scale language model server through a chat program by a resident program of the user terminal. [Figure 5] FIG. 10 is a diagram illustrating a configuration of an information processing system according to a second embodiment. [Figure 6] FIG. 10 is a block diagram showing an example of a hardware configuration of an image forming apparatus in a second embodiment. [Figure 7A] FIG. 10 is a diagram showing a database constructed in a large-scale language model server in the second embodiment. [Figure 7B] A flowchart showing the operation of the large-scale language model server in the second embodiment when the image forming device performs an operation on a file and the URL of the file is sent to the large-scale language model server by a resident program of the image forming device. [Figure 7C] 10 is a flowchart showing the operation of the large-scale language model server in the second embodiment when a question sentence and a user ID are transmitted to the large-scale language model server through a chat program by a resident program of the image forming apparatus. [Figure 7D] 10 is a flowchart showing the operation of the large-scale language model server in the second embodiment, when the large-scale language model server periodically checks whether a file exists at a URL in each data element of the database. [Figure 8A] 10 is a flowchart showing the operation of the image forming apparatus when an operation related to a file is performed on the image forming apparatus itself, a file server, or a user terminal in the second embodiment. [Figure 8B] FIG. 11 is a diagram showing a screen displayed on the operation unit of the image forming apparatus when the image forming apparatus transmits a question about the storage location of a file to a large-scale language model server in the second embodiment. [Figure 8C] 10 is a flowchart showing the operation of the image forming apparatus in the second embodiment when a question sentence and a user ID are transmitted to a large-scale language model server through a chat program by a resident program of the image forming apparatus. DETAILED DESCRIPTION OF THE INVENTION

[0011] Each embodiment of the present invention will be described in detail below with reference to the drawings. However, the configurations described in each of the following embodiments are merely examples, and the scope of the present invention is not limited to the configurations described in each embodiment. For example, each component constituting the present invention can be replaced with any configuration that can perform the same function. Also, any component may be added. Furthermore, any two or more configurations (features) of each embodiment can be combined.

[0012] First Embodiment An overview of the first embodiment is given below. In the first embodiment, when a user terminal, which is a PC, performs an operation on a file on a file server on an in-house intranet, a large-scale language model learns the contents of the file. Thereafter, the user terminal indicates the contents of the file and asks the large-scale language model, using natural language, a question about the storage location of the file. In response, the large-scale language model replies to the user terminal, using natural language, with the URL (Uniform Resource Locator) of the file that is the subject of the query.

[0013] In this specification, "file-related operations" refer to file reading, writing, new creation, deletion, execution, etc. Note that new creation includes creating a new file and storing a newly created file. Also, in this specification, "URL" includes a configuration consisting only of a file path within a device such as a user terminal. Hereinafter, a first embodiment will be described with reference to FIGS. 1 to 4C.

[0014] [System Configuration] Fig. 1 is a diagram showing the configuration of an information processing system 100 according to the first embodiment. As shown in Fig. 1, the information processing system 100 includes a user terminal 101, a file server 102, a large-scale language model server 103, a router 104, and a local area network 105 (hereinafter abbreviated as "network 105"). In the information processing system 100, the user terminal 101 (information processing device), the file server 102, and the large-scale language model server 103 (server) can communicate with each other via the router 104 and the network 105. Note that each of the user terminal 101 and the file server 102 may not be connected to the network 105 in a single configuration, but may be connected in multiple configurations.

[0015] The network 105 is a network constructed using cables such as Ethernet laid on each floor of the company. The router 104 is responsible for routing between the multiple networks 105. The router 104 can also use VPN technology to treat networks at physically separated company locations as the same network. The file server 102 is installed in each department or location within the company and is an SMB server that allows all employees or authorized users within the company to share files. Files in the file server 102 are uniquely identified by URLs. Therefore, the user terminal 101 and the large-scale language model server 103 can use URLs to perform operations related to files on the file server 102.

[0016] The user terminal 101 is a terminal that a user can connect to the network 105, such as a PC, a tablet terminal, or a smartphone. The user terminal 101 typically has software such as a browser, a word processor, and a spreadsheet, and can read, write, delete, execute, and so on, files in a file server 102 on the network 105 using this software. The user terminal 101 can also store newly created files in the file server 102 on the network 105 using the above-mentioned software. In addition, a resident program runs on the user terminal 101.

[0017] In the user terminal 101, a resident program detects file-related operations performed from the user terminal 101 on the file server 102. Furthermore, if the detected file-related operation is performed on a document file written in a natural language or an image file containing a natural language, the resident program transmits the URL of the file and the like to the large-scale language model in the large-scale language model server 103. Furthermore, a chat program that can communicate with the large-scale language model in natural language also runs on the user terminal 101.

[0018] The file server 102 has a function of providing files in response to requests from the user terminal 101 and the large-scale language model server 103. In this specification, the communication protocol used when the file server 102 provides files to the user terminal 101 and the large-scale language model server 103 is SMB, as described above. However, the communication protocol is not particularly limited, and may be, for example, HTTP (Hyper Text Transfer Protocol) or WebDAV, or another communication protocol.

[0019] In the large-scale language model server 103, the large-scale language model acquires the file of the URL transmitted from the user terminal 101. If the acquired file is a document file written in a natural language, the large-scale language model converts the content of the natural language into a language vector. If the acquired file is an image file, the large-scale language model extracts the natural language contained in the image using OCR or a machine learning method, and converts the extracted natural language content into a language vector. Furthermore, the large-scale language model builds a database in the large-scale language model server 103, whose data elements are combinations of at least URLs and language vectors. Furthermore, when a question in natural language regarding the storage location of a file is transmitted from the user terminal 101, the large-scale language model converts the question into a language vector and calculates the similarity between the language vector of the question and each language vector in the database. Furthermore, the large-scale language model responds to the user terminal 101 with the URL combined with the language vector with the highest calculated similarity.

[0020] [Hardware configuration] 2 is a block diagram showing an example of the hardware configuration of a computer 200. The user terminal 101, the file server 102, and the large-scale language model server 103 each have a computer mechanism and therefore share a common hardware configuration. Therefore, the hardware configuration common to the user terminal 101, the file server 102, and the large-scale language model server 103 will be described using the hardware configuration of the computer 200 shown in FIG. 2. As shown in FIG. 2, in the computer 200, a CPU 201, a ROM 202, a RAM 203, a HDD 204, an operation unit I / F 205, a display unit I / F 206, and a network I / F 207 are connected to a system bus 208.

[0021] The CPU 201 starts the OS by a boot program stored in the ROM 202. The CPU 201 executes various processes on the started OS by executing application programs stored in the HDD 204. The RAM 203 is used as the working area of the CPU 201. The HDD 204 stores application programs and data managed by the application programs. The network I / F 207 is connected to the network 105.

[0022] The application program inputs and outputs information to and from each device on the network 105 via the network 105. The operation unit I / F 205 is an interface with the operation unit 209, which is made up of a mouse, keyboard, etc., and sends operation information input by a user on the operation unit 209 to the CPU 201. The display unit I / F 206 outputs display screen data generated by the CPU 201 to a display unit 210, which is made up of a display, etc. The operation unit 209 and the display unit 210 implement a user interface.

[0023] [Details of the large-scale language model server] The large-scale language model server 103 will be described in detail below with reference to Figures 3A to 3C. Figure 3A is a diagram showing a database 300 constructed in the large-scale language model server 103. As shown in Figure 3A, the database 300 has a plurality of data elements 301. Each data element 301 includes three pieces of information: a URL indicating the storage location of a certain file, the user ID of the user terminal 101 where an operation related to the file was performed, and a language vector obtained by converting the content of the file into a language vector. Note that in the database 300, each time a URL is sent from the user terminal 101 to the large-scale language model in the large-scale language model server 103, a data element 301 is updated or a new data element 301 is added.

[0024] 3B and 3C are flowcharts showing the operation of the large-scale language model in the large-scale language model server 103. Note that, hereinafter, the large-scale language model in the large-scale language model server 103 may be abbreviated as the "large-scale language model server 103." FIG. 3B is a flowchart showing the operation of the large-scale language model server 103 when, when the user terminal 101 performs an operation related to a file, the URL of the file is sent to the large-scale language model server 103 by a resident program of the user terminal 101. The operation shown in the flowchart in FIG. 3B is realized in the large-scale language model server 103 by the CPU 201 expanding the large-scale language model program stored in the HDD 204 into the RAM 203 and executing it.

[0025] In step S301, the CPU 201 of the large-scale language model server 103 determines whether a URL and a user ID have been transmitted from the user terminal 101. If the CPU 201 of the large-scale language model server 103 determines that a URL and a user ID have not been transmitted from the user terminal 101, the process returns to step S301. As a result, the CPU 201 of the large-scale language model server 103 enters a state of waiting for a URL and a user ID to be transmitted from the user terminal 101. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that a URL and a user ID have been transmitted from the user terminal 101, the process proceeds to step S302. In step S302, the CPU 201 (acquisition means) of the large-scale language model server 103 acquires the file of the URL transmitted from the user terminal 101.

[0026] In step S303, the CPU 201 of the large-scale language model server 103 determines whether the acquired file is an image file. If the CPU 201 of the large-scale language model server 103 determines that the acquired file is an image file, the process proceeds to step S304. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the acquired file is not an image file, the process proceeds to step S305. In step S304, the CPU 201 of the large-scale language model server 103 extracts natural language from the acquired image file. In step S305, the CPU 201 of the large-scale language model server 103 converts the natural language included in the acquired file or the natural language extracted from the acquired file into a language vector. As a result, the CPU 201 of the large-scale language model server 103 acquires the language vector of the natural language included in the acquired file or the natural language extracted from the acquired file.

[0027] The processing from the next step onwards is the operation of the large-scale language model server 103 on the database 300. In step S306, the CPU 201 of the large-scale language model server 103 determines whether the URL transmitted from the user terminal 101 exists in the database 300. This determination is made by the CPU 201 of the large-scale language model server 103 searching for URLs included in all data elements 301 in the database 300. If the CPU 201 of the large-scale language model server 103 determines that the URL transmitted from the user terminal 101 does not exist in the database 300, the processing proceeds to step S307. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the URL transmitted from the user terminal 101 exists in the database 300, the processing proceeds to step S308.

[0028] In step S307, the CPU 201 (storage means) of the large-scale language model server 103 creates a new data element 301 in the database 300. At that time, the CPU 201 of the large-scale language model server 103 combines the language vector acquired in step S305 with the URL and user ID sent from the user terminal 101 and includes them in the new data element 301. Thereafter, the process returns to step S301. In step S308, the CPU 201 of the large-scale language model server 103 determines whether the language vector included in the data element 301 together with the URL sent from the user terminal 101 matches the language vector acquired in step S305.

[0029] If the CPU 201 of the large-scale language model server 103 determines that the language vector included in the data element 301 together with the URL transmitted from the user terminal 101 does not match the language vector acquired in step S305, the process proceeds to step S309. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the language vector included in the data element 301 together with the URL transmitted from the user terminal 101 matches the language vector acquired in step S305, the process proceeds to step S310. In step S309, the CPU 201 of the large-scale language model server 103 replaces the language vector combined in the URL transmitted from the user terminal 101 and the data element 301 with the language vector acquired in step S305.

[0030] Furthermore, if the user ID combined with the URL transmitted from the user terminal 101 in the data element 301 differs from the user ID transmitted from the user terminal 101 together with the URL, the CPU 201 of the large-scale language model server 103 adds the user ID. In the addition of the user ID, the user ID transmitted from the user terminal 101 together with the URL is added to the data element 301 containing the URL transmitted from the user terminal 101. Thereafter, the process returns to step S301. In step S310, the CPU 201 of the large-scale language model server 103 adds the user ID in the same manner as in step S309. Note that in step S310, the language vector replacement of step S309 is not performed. Thereafter, the process returns to step S301.

[0031] 3C is a flowchart showing the operation of the large-scale language model server 103 when a question and a user ID are sent to the large-scale language model server 103 via a chat program by a resident program of the user terminal 101. Note that the question in the first embodiment is a sentence written in natural language in which the user asks a question about the storage location of a file by indicating the contents of the file. The operation shown in the flowchart in FIG. 3C is realized in the large-scale language model server 103 by the CPU 201 expanding the large-scale language model program stored in the HDD 204 into the RAM 203 and executing it.

[0032] In step S311, the CPU 201 of the large-scale language model server 103 determines whether a question text and a user ID have been transmitted from the user terminal 101. If the CPU 201 of the large-scale language model server 103 determines that a question text and a user ID have not been transmitted from the user terminal 101, the process returns to step S311. This causes the CPU 201 of the large-scale language model server 103 to wait for a question text and a user ID to be transmitted from the user terminal 101. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that a question text and a user ID have been transmitted from the user terminal 101, the process proceeds to step S312.

[0033] In step S312, the CPU 201 of the large-scale language model server 103 converts the question sent from the user terminal 101 into a language vector. As a result, the CPU 201 of the large-scale language model server 103 acquires the language vector of the question sent from the user terminal 101. In step S313, the CPU 201 of the large-scale language model server 103 calculates the similarity between the language vector included in each data element 301 and the language vector acquired in step S312 for all data elements 301 in the database 300. Furthermore, the CPU 201 (search means) of the large-scale language model server 103 identifies the data element 301 with the greatest language vector similarity among all data elements 301 in the database 300. Note that the similarity may be calculated using any method, such as cosine similarity, as long as it is a method that can calculate the similarity between language vectors.

[0034] In step S314, the CPU 201 of the large-scale language model server 103 determines whether the user ID included in the data element 301 identified in step S313 matches the user ID sent from the user terminal 101. This determination ensures security for the search results of the storage locations of in-house document files using the user ID. If the CPU 201 of the large-scale language model server 103 determines that the user ID included in the data element 301 identified in step S313 matches the user ID sent from the user terminal 101, the process proceeds to step S315. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the user ID included in the data element 301 identified in step S313 does not match the user ID sent from the user terminal 101, the process proceeds to step S316.

[0035] In step S315, the CPU 201 (response sending means) of the large-scale language model server 103 sends a response sentence (response message) conveying the URL included in the data element 301 identified in step S313 to the user terminal 101. Note that the response sentence refers to a sentence in natural language provided by the large-scale language model in response to the user's question about the storage location of the file. Thereafter, the process returns to step S311. In step S316, the CPU 201 of the large-scale language model server 103 sends a response sentence (absence message) to the user terminal 101 indicating that there is no file corresponding to the question from the user terminal 101. Thereafter, the process returns to step S311.

[0036] [User terminal operation] The operation of the user terminal 101 will be described below with reference to Figures 4A to 4C. Figure 4A is a flowchart showing the operation of the user terminal 101 when the user terminal 101 performs a file-related operation on the user terminal 101 itself or on the file server 102. The operation shown in the flowchart of Figure 4A is realized in the user terminal 101 by the CPU 201 expanding a resident program stored in the HDD 204 into the RAM 203 and executing it.

[0037] In step S401, CPU 201 of user terminal 101 determines whether a file-related operation has been performed on user terminal 101 itself or file server 102. If CPU 201 of user terminal 101 determines that a file-related operation has been performed on user terminal 101 itself or file server 102, the process proceeds to step S402. On the other hand, if CPU 201 of user terminal 101 determines that a file-related operation has not been performed on either user terminal 101 itself or file server 102, the process returns to step S401. As a result, CPU 201 of user terminal 101 enters a state of waiting for a file-related operation to be performed on either user terminal 101 itself or file server 102.

[0038] In step S402, the CPU 201 of the user terminal 101 determines whether the file determined in step S401 contains a natural language. The method of determination in step S402 is not particularly limited, and may be a method of determining simply based on the file extension, a method of determining based on the presence or absence of a character code, or a method of determining the file type based on the presence of a specific binary string at the beginning of the file. If the CPU 201 of the user terminal 101 determines that the file determined in step S401 does not contain a natural language, the process returns to step S401. On the other hand, if the CPU 201 of the user terminal 101 determines that the file determined in step S401 does contain a natural language, the process proceeds to step S403.

[0039] In step S403, the CPU 201 (transmission means) of the user terminal 101 transmits the user ID and the URL of the file determined in step S401 to the large-scale language model server 103. Thereafter, the process returns to step S401. Note that the method of transmission to the large-scale language model server 103 is not particularly limited, and may be, for example, the HTTP POST method or the FTP PUT method. This also applies to the second embodiment described later.

[0040] 4B is a diagram showing a screen 400 that is displayed on the display unit 210 of the user terminal 101 when the user terminal 101 sends a question about the storage location of a file to the large-scale language model server 103. In the user terminal 101, the screen 400 is displayed on the display unit 210 by a chat program. An example of a text chat that takes place between a user operating the user terminal 101 and a large-scale language model in the large-scale language model server 103 will be described below using the screen 400 in FIG. 4B.

[0041] As shown on screen 400 in FIG. 4B, a user asks the large-scale language model a question about the storage location of Company A's invoice for April of this year through chat comment 401, in a conversational natural language style. In response to the question, the large-scale language model responds to the user through chat comment 402, in a conversational natural language style, by informing the user of the URL of a shared folder on file server 102 as the storage location of Company A's invoice for April of this year. Note that the URL in chat comment 402 is configured as a Windows (registered trademark) file path, but depending on the communication environment, it may be a URL with an SMB scheme. This also applies to the second embodiment described below.

[0042] The user also asks the large-scale language model a question about the storage location of the price list for Company B's tires for fall 2023, in a natural language, using a word-delimited sentence, through chat comment 403. In response to the question, the large-scale language model responds to the user through chat comment 404 in a conversational natural language, providing a URL on an HTTP server (not shown) as the storage location of Company B's tire price list for fall 2023. In chat comment 404, the communication protocol indicated by the URL scheme is HTTP.

[0043] Therefore, an HTTP server (not shown) provides files to the user terminal 101 and the large-scale language model server 103 via the network 105 using the HTTP communication protocol. The HTTP server (not shown) has the hardware configuration shown in FIG. 2, similar to the file server 102. In this way, a user operating the user terminal 101 can easily and quickly search for the storage location of an in-house document file by text chatting with the large-scale language model in the large-scale language model server 103. This also applies to the second embodiment described below.

[0044] 4C is a flowchart showing the operation of the user terminal 101 when a question sentence and a user ID are sent to the large-scale language model server 103 through a chat program by a resident program of the user terminal 101. The operation (control method of an information processing device) shown in the flowchart of FIG. 4C is realized in the user terminal 101 by the CPU 201 (computer) expanding a resident program (program) stored in the HDD 204 into the RAM 203 and executing it.

[0045] In step S411, the CPU 201 of the user terminal 101 determines whether a question has been input to the user terminal 101. If the CPU 201 of the user terminal 101 determines that a question has been input to the user terminal 101, the process proceeds to step S412. At that time, the CPU 201 of the user terminal 101 displays the question input to the user terminal 101 on the display unit 210, for example, as shown in the chat comments 401 and 403 on the screen 400 of FIG. 4B. On the other hand, if the CPU 201 of the user terminal 101 determines that a question has not been input to the user terminal 101, the process returns to step S411. This causes the CPU 201 of the user terminal 101 to wait for a question to be input to the user terminal 101. In step S412, the CPU 201 (message sending means) of the user terminal 101 sends the question (question message) and the user ID to the large-scale language model server 103 (message sending step).

[0046] In step S413, the CPU 201 of the user terminal 101 determines whether a response sentence has been sent from the large-scale language model server 103. If the CPU 201 of the user terminal 101 determines that a response sentence has not been sent from the large-scale language model server 103, the process returns to step S413. This causes the CPU 201 of the user terminal 101 to wait for a response sentence to be sent from the large-scale language model server 103. On the other hand, if the CPU 201 of the user terminal 101 determines that a response sentence has been sent from the large-scale language model server 103, the process proceeds to step S414. In step S414, the CPU 201 (message receiving means) of the user terminal 101 receives the response sentence from the large-scale language model server 103 (message receiving step). At this time, the CPU 201 of the user terminal 101 displays the response sentence from the large-scale language model server 103 on the display unit 210, for example, as chat comments 402 and 404 on the screen 400 of FIG. 4B. Then, the process returns to step S411.

[0047] [summary] As described above, in the information processing system 100, the resident program of the user terminal 101 detects that a file-related operation has occurred on the file server 102. At that time, the resident program transmits the URL of the file related to the detected operation to the large-scale language model server 103. In the large-scale language model server 103, the large-scale language model acquires the file of the URL transmitted from the user terminal. Furthermore, the large-scale language model constructs a database 300 having data elements 301 that combine the language vector of the natural language contained in the acquired file with the URL.

[0048] Thereafter, the user sends a natural language question from the user terminal 101 to the large-scale language model via text chat to search for the storage location of the file. In this case, the large-scale language model calculates the similarity between the language vector of the question and the language vector in each data element 301 in the database 300. Furthermore, the large-scale language model sends a response message in natural language to the user terminal 101 via text chat, conveying the URL in the data element 301 with the highest calculated similarity. In response, the user terminal 101 receives the response message from the large-scale language model and displays the received response message on the display unit 210, thereby notifying the user of the search results that best fit the question.

[0049] In this way, the information processing system 100 and the user terminal 101 can easily search for a URL, which is information indicating the storage location of a file, without being restricted by how the file was stored on the network 105.

[0050] Second Embodiment An overview of the second embodiment is given below. In the second embodiment, when an image forming apparatus performs an operation related to a file on a file server on an in-house intranet, a large-scale language model learns the contents of the file. In this case, the large-scale language model learns not only the contents of the file, but also metadata such as when, by whom, and what operation was performed on the file.

[0051] Recently, corporate intranets are often designed to use a common user ID across all devices within a company, including different hosts, i.e., image forming devices, file servers, and PCs. For example, when a user performs card authentication on an image forming device and then sends a scanned file of an internal document to a file server, the user can create a new file on the file server with the same user privileges as the PC. This allows both the image forming device and the PC to prompt the large-scale language model to add a new data element to the database. Furthermore, a device on a network different from the image forming device or PC that prompted the addition of the new data element can ask the large-scale language model questions about the file's storage location using natural language.

[0052] Furthermore, the large-scale language model periodically, for example, every 12 hours, checks whether a file exists at the URL in each data element of the database. For a URL where a file does not exist, the large-scale language model checks whether a file with the same content as the file exists in a directory near the URL. If the large-scale language model finds a file with the same content as the file in a directory near the URL where a file does not exist, it replaces the URL where the file does not exist in the data element of the database with the URL of the file with the same content. The second embodiment will be described below with reference to Figures 5 to 8C. Here, differences from the first embodiment will be mainly described. Note that in the second embodiment, the same configurations and processes as those in the first embodiment are denoted by the same reference numerals as those in the first embodiment, and detailed description thereof will be omitted.

[0053] [System Configuration] FIG. 5 is a diagram showing the configuration of an information processing system 500 according to the second embodiment. As shown in FIG. 5, the information processing system 500 according to the second embodiment includes a user terminal 101, a file server 102, a large-scale language model server 103, a router 104, a network 105, and an image forming apparatus 501 (information processing apparatus). In the information processing system 500, the user terminal 101, the file server 102, the large-scale language model server 103, and the image forming apparatus 501 can communicate with each other via the router 104 and the network 105. Therefore, the information processing system 500 according to the second embodiment is the information processing system 100 according to the first embodiment, with the image forming apparatus 501 connected to the network 105. Therefore, detailed descriptions of the configuration of the information processing system 500 according to the second embodiment and the hardware configuration of the user terminal 101 and the like will be omitted, and the hardware configuration of the image forming apparatus 501 will be described in detail below.

[0054] [Hardware configuration of image forming device] FIG. 6 is a diagram illustrating an example of the hardware configuration of the image forming apparatus 501. As shown in FIG.

[0055] As shown in Fig. 6, the image forming apparatus 501 has a controller unit 600. A scanner 601, which is an image input device, a printer 602, which is an image output device, and an operation unit 603 are connected to the controller unit 600. The controller unit 600 performs control to realize a copy function in which image data read by the scanner 601 is printed out by the printer 602. The controller unit 600 has a CPU 604. The CPU 604 starts the OS by a boot program stored in a ROM 605. The CPU 604 executes various processes on the started OS by executing programs stored in a storage 606.

[0056] The ROM 605 also stores setting values for various settings of the image forming apparatus 501. These setting values include an inference flag that indicates whether the image forming apparatus 501 operates in inference mode or learning mode. The RAM 607 is used as a work area for the CPU 604. The RAM 607 provides a work area as well as an image memory area for temporarily storing image data. The storage 606 stores programs, image data, and information about the paper feed capacity, including the maximum number of sheets that can be fed by the image forming apparatus 501.

[0057] The CPU 604 is connected to other modules via a system bus 608. The operation unit I / F 609 is an interface with the operation unit 603 having a touch panel, and outputs image data to be displayed on the operation unit 603 to the operation unit 603. The operation unit I / F 609 also sends information input by a user via the operation unit 603 to the CPU 604. The network I / F 610 is an interface for connecting the image forming apparatus 501 to the network 105. The USB host I / F 611 is an interface for communicating with a USB storage 612. The USB host I / F 611 is also an output unit for storing data stored in the storage 606 in the USB storage 612. The USB host I / F 611 also inputs data stored in the USB storage 612 and sends the input data to the CPU 604. The USB storage 612 is an external storage device that stores data and is detachable from the USB host I / F 611.

[0058] An authentication unit 613 can also be connected to the USB host I / F 611. A typical example of the authentication unit 613 is a card reader. When a user holds their authentication card over the card reader, which is the authentication unit 613, subsequent operations on the image forming apparatus 501 are considered to have been performed by the person identified by the user ID stored on the authentication card held over the card reader. As described above, the authentication card owned by the user stores a user ID that is common to all devices in the company, but a different user ID may be stored for each device or each user. Therefore, the authentication unit 613 can be a card reader or a device that performs biometric authentication using fingerprints, veins, or the like. In this way, multiple USB devices, including the USB storage 612 and the authentication unit 613, can be connected to the USB host I / F 611.

[0059] The RTC 614 controls time information. The time information controlled by the RTC 614 is used to obtain the time when auto-shutdown is performed, record the time when a job is submitted, and so on. The image bus I / F 615 is a bus bridge that connects the system bus 608 to an image bus 616 that transfers image data at high speed and converts data formats. The image bus 616 is configured using a PCI bus, IEEE 1394, or the like. The image bus 616 is connected to a device I / F 617, a scanner image processing unit 618, and a printer image processing unit 619. The device I / F 617 is connected to a scanner 601 and a printer 602. The device I / F 617 converts image data between synchronous and asynchronous systems. The scanner image processing unit 618 corrects, processes, and edits input image data. The printer image processing unit 619 corrects, converts resolution, and so on for print output image data according to the printer 602.

[0060] [Details of the large-scale language model server] The large-scale language model server 103 will be described in detail below with reference to FIGS. 7A to 7D. FIG. 7A is a diagram showing a database 700 constructed in the large-scale language model server 103. As shown in FIG. 7A, the database 700 has a plurality of data elements 701. Each data element 701 includes four pieces of information. The four pieces of information are a URL indicating the storage location of a certain file, the user ID of the device on which an operation related to the file was performed, a language vector obtained by converting the content of the file into a language vector, and metadata at the time of new creation of the file. In this respect, compared to each data element 301 in the first embodiment, each data element 701 differs in that it includes a language vector obtained by converting metadata at the time of new creation of the file into a language vector, and metadata at the time of new creation of the file. The metadata at the time of new creation of the file includes the creation date and time, the name of the creator, and the content of the action (cause of creation) at the time of new creation of the file. In the database 700, each time a URL is transmitted from the image forming device 501 or the like to the large-scale language model in the large-scale language model server 103, a data element 701 is updated or a new data element 701 is added.

[0061] 7B to 7D are flowcharts showing the operation of the large-scale language model in the large-scale language model server 103. Fig. 7B is a flowchart showing the operation of the large-scale language model server 103 when the image forming device 501 performs an operation related to a file and the URL of the file is sent to the large-scale language model server 103 by a resident program of the image forming device 501. The operation shown in the flowchart in Fig. 7B is realized in the large-scale language model server 103 by the CPU 201 expanding the program of the large-scale language model stored in the HDD 204 into the RAM 203 and executing it.

[0062] In step S701, the CPU 201 of the large-scale language model server 103 determines whether a URL and a user ID have been sent from the image forming device 501. Furthermore, if the file-related operation on the image forming device 501 is an operation to create a new file, the CPU 201 of the large-scale language model server 103 also determines whether metadata at the time of new file creation has been sent from the image forming device 501. Note that when the file-related operation on the image forming device 501 is an operation to create a new file, for example, there is an operation in which a user sends and stores a file created by scanning an in-house document on the image forming device 501 to the file server 102.

[0063] If the CPU 201 of the large-scale language model server 103 determines that the URL, user ID, etc. have not been transmitted from the image forming device 501, the process returns to step S701. This puts the CPU 201 of the large-scale language model server 103 into a state of waiting for the URL, user ID, etc. to be transmitted from the image forming device 501. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the URL, user ID, etc. have been transmitted from the image forming device 501, the process proceeds to step S302. Steps S302, S303, and S304 are the same as those in the first embodiment, and therefore detailed description thereof will be omitted.

[0064] In step S702, the CPU 201 of the large-scale language model server 103 converts the natural language included in the acquired file or the natural language extracted from the acquired file into a language vector. Furthermore, if metadata has been transmitted from the image forming device 501, the CPU 201 of the large-scale language model server 103 also converts the metadata into a language vector. As a result, the CPU 201 of the large-scale language model server 103 acquires the language vector of the natural language included in the acquired file or the natural language extracted from the acquired file. Furthermore, if metadata has been transmitted from the image forming device 501, the CPU 201 of the large-scale language model server 103 also acquires the language vector of the metadata.

[0065] In step S703, the CPU 201 of the large-scale language model server 103 determines whether the URL transmitted from the image forming device 501 exists in the database 700. This determination is made in the same manner as the determination in step S306 in the first embodiment. If the CPU 201 of the large-scale language model server 103 determines that the URL transmitted from the image forming device 501 does not exist in the database 700, the process proceeds to step S704. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the URL transmitted from the image forming device 501 exists in the database 700, the process proceeds to step S705.

[0066] In step S704, the CPU 201 (storage means) of the large-scale language model server 103 creates a new data element 701 in the database 700. At this time, the CPU 201 of the large-scale language model server 103 combines the language vector acquired in step S702 with the URL and user ID sent from the image forming device 501, and includes them in the new data element 701. Furthermore, if metadata has been sent from the image forming device 501, the CPU 201 of the large-scale language model server 103 also combines the metadata and includes it in the new data element 701. Thereafter, the process returns to step S701.

[0067] In step S705, the CPU 201 of the large-scale language model server 103 determines whether the language vector included in the data element 701 together with the URL transmitted from the image forming apparatus 501 matches the language vector acquired in step S702. If the CPU 201 of the large-scale language model server 103 determines that the language vector included in the data element 701 together with the URL transmitted from the image forming apparatus 501 does not match the language vector acquired in step S702, the process proceeds to step S706. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the language vector included in the data element 701 together with the URL transmitted from the image forming apparatus 501 matches the language vector acquired in step S702, the process proceeds to step S707.

[0068] In step S706, the CPU 201 of the large-scale language model server 103 replaces the language vector combined with the URL transmitted from the image forming apparatus 501 in the data element 701 with the language vector acquired in step S702. Furthermore, if the user ID combined with the URL transmitted from the image forming apparatus 501 in the data element 701 differs from the user ID transmitted from the image forming apparatus 501 together with the URL, the CPU 201 of the large-scale language model server 103 adds the user ID. To add the user ID, the user ID transmitted from the image forming apparatus 501 together with the URL is added to the data element 701 containing the URL transmitted from the image forming apparatus 501. Then, the process returns to step S701. In step S707, the CPU 201 of the large-scale language model server 103 adds the user ID in the same manner as in step S706. Note that in step S707, the language vector replacement of step S706 is not performed. Then, the process returns to step S701.

[0069] FIG. 7C is a flowchart showing the operation of the large-scale language model server 103 when a question and a user ID are sent to the large-scale language model server 103 via a chat program by a resident program of the image forming device 501. The question in the second embodiment refers to a sentence written in natural language, such as a sentence in which a user asks a question about the storage location of a file by indicating the contents of the file, or a sentence in which the user indicates the creation date and time, the creator's name, and the action taken when creating a new file. The operation shown in the flowchart of FIG. 7C is realized by the CPU 201 of the large-scale language model server 103 loading the large-scale language model program stored in the HDD 204 into the RAM 203 and executing it. The explanation of steps S311 to S316 in FIG. 3C above becomes the explanation of steps S711 to S716 in FIG. 7C when the user terminal 101, database 300, and data element 301 are changed to the image forming device 501, database 700, and data element 701. Therefore, the operations shown in the flowchart of FIG. 7C are the same as the operations shown in the flowchart of FIG. 3C in the first embodiment, except for the points mentioned above, and therefore detailed description thereof will be omitted.

[0070] 7D is a flowchart showing the operation of the large-scale language model server 103 when the large-scale language model server 103 periodically checks whether a file exists at the URL in each data element 701 of the database 700. The large-scale language model server 103 periodically checks every 12 hours whether a file exists at the URL in each data element 701 of the database 700. Furthermore, if the file to be checked at the URL in the data element 701 does not exist or is not located near the URL, the large-scale language model server 103 deletes the data element 701 from the database 700. Furthermore, if the file to be checked at the URL exists in a directory near the URL in the data element 701, the large-scale language model server 103 replaces the URL in the data element 701 with the URL where the file to be checked exists. Note that in the second embodiment, the directory near the URL refers to the parent directory of the directory of the URL or a directory at the same level as the directory of the URL, but is not limited thereto.

[0071] The operation shown in the flowchart of FIG. 7D is realized by the CPU 201 of the large-scale language model server 103 by loading a program stored in the HDD 204 into the RAM 203 and executing it. In step S721, the CPU 201 of the large-scale language model server 103 determines whether 12 hours have passed since startup or the last URL check. The CPU 201 of the large-scale language model server 103 measures 12 hours using an internal clock. If the CPU 201 of the large-scale language model server 103 determines that 12 hours have not passed since startup or the last URL check, the process returns to step S721. This puts the CPU 201 of the large-scale language model server 103 into a state of waiting for 12 hours to pass since startup or the last URL check. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that 12 hours have passed since startup or the last URL check, the process proceeds to step S722.

[0072] In step S722, the CPU 201 of the large-scale language model server 103 sets the first data element 701 as the target of URL check. In step S723, the CPU 201 (determination means) of the large-scale language model server 103 determines whether a file exists at the URL included in the data element 701 targeted for URL check. This determination is made based on whether the CPU 201 of the large-scale language model server 103 has reached a file at the URL included in the data element 701 targeted for URL check. If the CPU 201 of the large-scale language model server 103 determines that no file exists at the URL included in the data element 701 targeted for URL check, the process proceeds to step S724. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that a file exists at the URL included in the data element 701 targeted for URL check, the process proceeds to step S725.

[0073] In step S724, the CPU 201 (determination means) of the large-scale language model server 103 determines whether a file with the same file name exists in a directory near the URL included in the data element 701 whose URL has been checked. The file name is included in the URL. If the CPU 201 of the large-scale language model server 103 determines that a file with the same file name exists in a directory near the URL included in the data element 701 whose URL has been checked, the process proceeds to step S725. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that a file with the same file name does not exist in a directory near the URL included in the data element 701 whose URL has been checked, the process proceeds to step S726.

[0074] In step S725, the CPU 201 of the large-scale language model server 103 determines whether the language vector of the file determined in step S724 matches the language vector included in the data element 701 whose URL was checked. In this case, the CPU 201 of the large-scale language model server 103 acquires the file determined in step S724 and converts the natural language included in the acquired file into a language vector together with the metadata of the file. However, if the file determined in step S724 is an image file, the CPU 201 of the large-scale language model server 103 converts the natural language extracted from the file into a language vector together with the metadata of the file. In this way, the CPU 201 of the large-scale language model server 103 acquires the language vector of the file determined in step S724.

[0075] If the CPU 201 of the large-scale language model server 103 determines that the language vector of the file determined in step S724 does not match the language vector included in the data element 701 that was the target of URL confirmation, the process proceeds to step S726. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the language vector of the file determined in step S724 matches the language vector included in the data element 701 that was the target of URL confirmation, the process proceeds to step S727. In step S726, the CPU 201 (deletion means) of the large-scale language model server 103 deletes the data element 701 that was the target of URL confirmation from the database 700. As a result, the unnecessary data element 701 is erased from the database 700. Thereafter, the process proceeds to step S729.

[0076] In step S727, the CPU 201 of the large-scale language model server 103 determines whether the determination in step S724 has been made for the data element 701 whose URL was checked. If the CPU 201 of the large-scale language model server 103 determines that the determination in step S724 has been made for the data element 701 whose URL was checked, the process proceeds to step S728. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the determination in step S724 has not been made for the data element 701 whose URL was checked, the process proceeds to step S729. In step S728, the CPU 201 (information replacement means) of the large-scale language model server 103 replaces the URL of the data element 701 whose URL was checked with the URL of the file whose URL was checked in step S724. As a result, useless URLs in the data elements 701 of the database 700 are rewritten into useful URLs. If the determination in step S724 has not been made for the data element 701 whose URL has been checked, the URL of the data element 701 whose URL has been checked will not be replaced.

[0077] In step S729, the CPU 201 of the large-scale language model server 103 determines whether the data element 701 targeted for URL check is the last data element 701 in the database 700. If the CPU 201 of the large-scale language model server 103 determines that the data element 701 targeted for URL check is the last data element 701 in the database 700, the process returns to step S721. This puts the CPU 201 of the large-scale language model server 103 into a standby state until 12 hours have elapsed since the previous URL check. On the other hand, if the CPU 201 of the large-scale language model server 103 determines that the data element 701 targeted for URL check is not the last data element 701 in the database 700, the process proceeds to step S730. In step S730, the CPU 201 of the large-scale language model server 103 sets the next data element 701 as the target for URL check. Thereafter, the process returns to step S723.

[0078] [Operation of image forming device] The operation of image forming apparatus 501 will be described below with reference to Figures 8A to 8C. Figure 8A is a flowchart showing the operation of image forming apparatus 501 when image forming apparatus 501 performs a file-related operation on image forming apparatus 501 itself, file server 102, or user terminal 101. The operation shown in the flowchart of Figure 8A is realized in image forming apparatus 501 by CPU 604 loading a resident program stored in storage 606 into RAM 607 and executing the program.

[0079] In step S801, the CPU 604 of the image forming apparatus 501 determines whether a file-related operation has been performed on the image forming apparatus 501 itself, the file server 102, or the user terminal 101. If the CPU 604 of the image forming apparatus 501 determines that a file-related operation has been performed on any of the image forming apparatus 501 itself, the file server 102, or the user terminal 101, the process proceeds to step S802. On the other hand, if the CPU 604 of the image forming apparatus 501 determines that a file-related operation has not been performed on any of the image forming apparatus 501 itself, the file server 102, and the user terminal 101, the process returns to step S801. As a result, the CPU 604 of the image forming apparatus 501 enters a state of waiting for a file-related operation to be performed on any of the image forming apparatus 501 itself, the file server 102, or the user terminal 101.

[0080] In step S802, the CPU 604 of the image forming apparatus 501 determines whether the file determined in step S801 contains a natural language. The determination method in step S802 is the same as the determination method in step S402 in the first embodiment. If the CPU 604 of the image forming apparatus 501 determines that the file determined in step S801 does not contain a natural language, the process returns to step S801. On the other hand, if the CPU 604 of the image forming apparatus 501 determines that the file determined in step S801 contains a natural language, the process proceeds to step S803.

[0081] In step S803, the CPU 604 of the image forming apparatus 501 determines whether the file determined in step S801 is a newly created file. If the CPU 604 of the image forming apparatus 501 determines that the file determined in step S801 is not a newly created file, the process proceeds to step S804. On the other hand, if the CPU 604 of the image forming apparatus 501 determines that the file determined in step S801 is a newly created file, the process proceeds to step S805.

[0082] In step S804, the CPU 604 (transmission means) of the image forming apparatus 501 transmits the user ID and the URL of the file determined in step S801 to the large-scale language model server 103. Thereafter, the process returns to step S801. In step S805, the CPU 604 (transmission means) of the image forming apparatus 501 transmits the user ID, the URL of the file determined in step S801, and metadata of the file when it was newly created to the large-scale language model server 103. Thereafter, the process returns to step S801.

[0083] 8B is a diagram showing a screen 800 displayed on the operation unit 603 of the image forming device 501 when the image forming device 501 transmits a question about the storage location of a file to the large-scale language model server 103. In the image forming device 501, the screen 800 is displayed on the operation unit 603 by a chat program. An example of a text chat conducted between a user of the image forming device 501 and the large-scale language model of the large-scale language model server 103 will be described below using the screen 800 of FIG. 8B. The question displayed on the screen 800 contains not only the content of the file but also the circumstances under which the file was created, or the content of the file plus the circumstances under which the file was created. This differs from the question displayed on the screen 400 of the first embodiment.

[0084] As shown on screen 800 in Figure 8B, a user asks the large-scale language model a question in a conversational style of natural language through chat comment 801 about the storage location of Company A's invoice for April of this year. However, the question only asks about the creation date, creator's name, and action taken when creating a new file. In response to the question, the large-scale language model responds to the user through chat comment 402 in a conversational style of natural language, informing them of the URL of a shared folder on file server 102 as the storage location of Company A's invoice for April of this year.

[0085] Furthermore, the user asks the large-scale language model, via chat comment 802, a question in a conversational natural language style about the storage location of the price list for Company B's tires for fall 2023. However, the question asks about the contents of the file, as well as the creation time and action taken when creating a new file. In response to the question, the large-scale language model responds to the user, via chat comment 404, in a conversational natural language style, by informing the user of a URL on an HTTP server (not shown) as the storage location of the price list for Company B's tires for fall 2023. Note that in chat comment 404, the communication protocol indicated by the URL scheme is HTTP. Therefore, the HTTP server (not shown) provides files to the image forming apparatus 501, the large-scale language model server 103, and the user terminal 101 via the network 105 using the HTTP communication protocol. Furthermore, the HTTP server (not shown) has the hardware configuration shown in FIG. 2, similar to the file server 102.

[0086] FIG. 8C is a flowchart showing the operation of the image forming apparatus 501 when a question sentence and a user ID are transmitted to the large-scale language model server 103 via a chat program by a resident program of the image forming apparatus 501. The operation (control method of an information processing apparatus) shown in the flowchart of FIG. 8C is realized by the image forming apparatus 501 having the CPU 604 (computer) load a resident program (program) stored in the storage 606 into the RAM 607 and execute it. The description of FIG. 4C above becomes the description of FIG. 8C when the user terminal 101, the CPU 201, the screen 400, and the chat comments 401 and 403 are changed to the image forming apparatus 501, the CPU 604, the screen 800, and the chat comments 801 and 802. Therefore, the operation shown in steps S811 to S814 of FIG. 8C is the same as the operation shown in steps S411 to S414 of FIG. 4C in the first embodiment except for the above-mentioned points, and therefore a detailed description thereof will be omitted.

[0087] [summary] As described above, in the second embodiment, the information processing system 500 and the image forming apparatus 501 can easily search for a URL, which is information indicating the storage location of a file, without being limited by the circumstances under which the file was stored on the network 105. As a result, when a user wants to print an in-house document that was scanned in the past, for example, in the image forming apparatus 501, the user can search for a URL indicating the storage location of the in-house document file in natural language. In this case, the user can search for a URL indicating the storage location of the file not only by the contents of the file but also by a question including the creation date, creator name, action taken when creating a new file, and the like.

[0088] <Other> Although the preferred embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications and changes are possible within the scope of the gist of the present invention.

[0089] For example, in the second embodiment, the image forming device 501 is an MFP (Multifunction Peripheral) that integrates a scan function, a copy function, and a communication function, but other functions may also be integrated. Furthermore, the image forming device 501 may also integrate a scan function and a communication function. Furthermore, in the second embodiment, the metadata at the time of new file creation is the creation date and time, the creator's name, and the action at the time of new file creation, but at least one of these is sufficient, and other information may also be included.

[0090] In the second embodiment, the user terminal 101 can also search for a URL indicating a storage location of a file in the same way as the image forming apparatus 501. Therefore, the user terminal 101 can search for a file that the user terminal 101 itself has stored in a device on the network 105, and the image forming apparatus 501 can also search for a file that the image forming apparatus 501 has stored in a device on the network 105. The flowchart of Fig. 7D in the second embodiment can also be applied to the first embodiment. The invention related to metadata in the second embodiment can also be applied to the first embodiment.

[0091] The present invention can also be realized by supplying a program that realizes one or more functions of each of the above embodiments to a system or device via a network or storage medium, and having one or more processors in the computer of the system or device read and execute the program. The present invention can also be realized by a circuit (e.g., ASIC) that realizes one or more functions.

[0092] The disclosure of each embodiment includes the following configurations, methods, and programs. (Configuration 1) An information processing device that can be connected via a network to a server on which a large-scale language model operates, which searches for information indicating file storage locations based on the similarity of language vectors of natural languages, a message sending means for sending a query message in a natural language to the large-scale language model to inquire about the storage location of the file; and a message receiving means for receiving from the large-scale language model a response message conveying information indicating the storage location of the file searched for by the large-scale language model using the language vector of the natural language of the query message. (Configuration 2) The information processing device according to configuration 1, wherein the message sending means and the message receiving means are text chats performed with the large-scale language model. (Configuration 3) The information processing device according to configuration 1, further comprising a transmitting means for transmitting information indicating the storage location of the file to the large-scale language model when an operation related to the file is performed on a device connected to the network. (Configuration 4) The information processing device according to Configuration 3, wherein the transmitting means transmits information indicating the storage location of the file to the large-scale language model only if the file contains natural language. (Configuration 5) The transmitting means transmits a user ID to the large-scale language model together with information indicating a storage location of the file; the message sending means sends the user ID together with the question message to the large-scale language model; The information processing device according to configuration 3 or 4, wherein the message receiving means receives the response message when a user ID sent to the large-scale language model together with information indicating the storage location of the file searched for by the large-scale language model using a language vector of the natural language of the question message matches a user ID sent to the large-scale language model together with the question message. (Configuration 6) The information processing device described in Configuration 5 is characterized in that the message receiving means receives an absence message indicating that the file does not exist instead of the response message when the user ID sent to the large-scale language model together with information indicating the storage location of the file searched by the large-scale language model using the language vector of the natural language of the question message does not match the user ID sent to the large-scale language model together with the question message. (Configuration 7) An information processing device described in any one of configurations 3 to 6, characterized in that the sending means, when the operation related to the file is a new creation operation, sends the metadata of the file to the large-scale language model together with information indicating the storage location of the file, in order to have the large-scale language model perform a search using a language vector of the metadata of the file. (Configuration 8) The information processing device according to configuration 7, wherein the metadata of the file is at least one of the creation date and time of the file, the name of the creator, and the creation factor. (Configuration 9) The information processing device according to any one of configurations 1 to 8, wherein the information indicating the storage location of the file is a URL (Uniform Resource Locator). (Configuration 10) The information processing device according to configuration 9, wherein the communication protocol indicated by the scheme of the URL is HTTP (Hyper Text Transfer Protocol). (Configuration 11) The information processing device according to configuration 9, wherein the communication protocol indicated by the scheme of the URL is SMB (Server Message Block). (Configuration 12) The information processing device according to any one of configurations 1 to 11, wherein the information processing device is a PC (Personal Computer). (Configuration 13) The information processing device according to any one of configurations 1 to 11, wherein the information processing device is an image forming device. (Configuration 14) The information processing device according to any one of configurations 1 to 13, wherein the file is a document file. (Configuration 15) The information processing device according to any one of configurations 1 to 13, wherein the file is an image file containing natural language. (Method 1) A control method for an information processing device that can be connected via a network to a server running a large-scale language model that searches for information indicating file storage locations based on similarities between language vectors of natural languages, the method comprising: a message sending step of sending a query message in a natural language to the large-scale language model to inquire about the storage location of the file; a message receiving step of receiving from the large-scale language model a response message conveying information indicating the storage location of the file searched for by the large-scale language model using a language vector of the natural language of the query message. (Program 1) A program for causing a computer to execute each means of the information processing device according to any one of configurations 1 to 15. (Configuration 16) An information processing system in which a server on which a large-scale language model runs and an information processing device can be connected via a network, The large-scale language model is an acquisition means for acquiring the file based on information indicating the storage location of the file; a storage means for storing in a database a data element that combines a language vector of the natural language included in the file and information indicating the storage location of the file; a search means for searching the database for information indicating a storage location of the file based on a similarity between a language vector of a natural language included in a query message from the information processing device and a language vector of a natural language included in the file; a response sending means for sending a response message to the information processing device to convey information indicating the storage location of the file searched for by the search means, The information processing device includes: a message sending means for sending a message inquiring about the storage location of the file in a natural language to the large-scale language model as the query message; and a message receiving means for receiving the response message from the large-scale language model. (Configuration 17) The information processing system according to configuration 16, wherein the message sending means and the message receiving means are text chats between the information processing device and the large-scale language model. (Configuration 18) The information processing system described in Configuration 16 or 17, characterized in that the information processing device includes a transmitting means for transmitting information indicating the storage location of the file to the large-scale language model when an operation related to the file is performed on a device connected to the network. (Configuration 19) In the information processing device, the information processing system described in Configuration 18 is characterized in that the transmitting means transmits information indicating the storage location of the file to the large-scale language model only if the file contains natural language. (Configuration 20) In the information processing device, the transmitting means transmits a user ID to the large-scale language model together with information indicating a storage location of the file; the message sending means sends the user ID together with the question message to the large-scale language model; In the large-scale language model, the storage means includes, in the database, the user ID transmitted from the information processing device to the large-scale language model together with the information indicating the storage location of the file, in a data element of the information indicating the storage location of the file; The information processing system described in configuration 18 or 19, wherein the response sending means sends the response message to the information processing device when a user ID included in a data element of information indicating the storage location of the file searched by the search means using a language vector of the natural language of the question message matches a user ID sent from the information processing device to the large-scale language model together with the question message. (Configuration 21) In the large-scale language model, the response sending means sends an absence message indicating that the file does not exist instead of the response message when the user ID included in the data element of the information indicating the storage location of the file searched by the search means using the language vector of the natural language of the question message does not match the user ID sent from the information processing device to the large-scale language model together with the question message. This is an information processing system described in Configuration 20. (Configuration 22) The information processing system described in Configuration 20 or 21, characterized in that the large-scale language model is provided with an addition means for adding, when a user ID sent from the information processing device to the large-scale language model together with information indicating the storage location of the file differs from a user ID included in a data element of the information indicating the storage location of the file, the user ID sent from the information processing device to the large-scale language model together with information indicating the storage location of the file to a data element of the information indicating the storage location of the file. (Configuration 23) In the information processing device, the transmitting means transmits metadata of the file together with information indicating a storage location of the file to the large-scale language model when the operation related to the file is a new file creation operation; In the large-scale language model, the storage means includes, in the database, a language vector of the metadata of the file in a data element of information indicating a storage location of the file; The information processing system according to any one of configurations 18 to 22, wherein the search means searches for information indicating the storage location of the file based on the similarity between a language vector of the natural language included in the query message from the information processing device and a language vector included in a data element of the database. (Configuration 24) The information processing system according to configuration 23, wherein the metadata of the file is at least one of the creation date and time of the file, the name of the creator, and the creation factor. (Configuration 25) The information processing system described in Configuration 23 or 24, characterized in that the large-scale language model includes a first vector replacement means that, when the language vectors of the natural language and file metadata contained in the file acquired by the acquisition means differ from the language vectors contained in the data elements of the information indicating the storage location of the file acquired by the acquisition means, replaces the language vectors contained in the data elements of the information indicating the storage location of the file acquired by the acquisition means with the language vectors of the natural language and file metadata contained in the file acquired by the acquisition means. (Configuration 26) An information processing system described in any one of configurations 16 to 22, characterized in that the large-scale language model includes a second vector replacement means that, when the language vector of the natural language included in the file acquired by the acquisition means is different from the language vector included in the data element of the information indicating the storage location of the file acquired by the acquisition means, replaces the language vector included in the data element of the information indicating the storage location of the file acquired by the acquisition means with the language vector of the natural language included in the file acquired by the acquisition means. (Configuration 27) The large-scale language model is a determination means for determining whether the file exists in the storage location of the file or in the vicinity of the storage location of the file based on information indicating the storage location of the file; and a deletion means for deleting a data element including information indicating the storage location of the file from the database when the file does not exist in the storage location of the file or in the vicinity of the storage location of the file. (Configuration 28) In the large-scale language model, the deletion means deletes a data element containing information indicating the storage location of the file from the database when the language vector for the file located at or near the storage location of the file differs from the language vector combined with information indicating the storage location of the file in a data element of the database. (Configuration 29) The information processing system described in Configuration 27 or 28 is characterized in that the large-scale language model is provided with a replacement means that, when a language vector for the file located near the storage location of the file matches a language vector combined with information indicating the storage location of the file in a data element of the database, in the data element of the database containing information indicating the storage location of the file, replaces the information indicating the storage location of the file with information indicating the vicinity of the storage location of the file. (Configuration 30) The information processing system according to configuration 29, wherein the information indicating the vicinity of the storage location of the file is a URL (Uniform Resource Locator). (Configuration 31) The information processing system according to any one of configurations 16 to 30, wherein the information indicating the storage location of the file is a URL. (Configuration 32) The information processing system according to configuration 30 or 31, wherein the communication protocol indicated by the scheme of the URL is HTTP (Hyper Text Transfer Protocol). (Configuration 33) The information processing system according to configuration 30 or 31, wherein the communication protocol indicated by the scheme of the URL is SMB (Server Message Block). (Configuration 34) The information processing system according to any one of configurations 16 to 33, wherein the information processing device is a PC (Personal Computer). (Configuration 35) The information processing system according to any one of configurations 16 to 33, wherein the information processing device is an image forming device. (Configuration 36) The information processing system according to any one of configurations 16 to 35, wherein the file is a document file. (Configuration 37) The information processing system according to any one of configurations 16 to 35, wherein the file is an image file containing natural language. [Explanation of symbols]

[0093] 101 User terminal (information processing device) 103 Large-scale language model server (server) 105 Network 201 CPU (message sending means) (message receiving means) 501 Image forming apparatus (information processing apparatus) 604 CPU (message sending means) (message receiving means)

Claims

1. An information processing device that can be connected via a network to a server on which a large-scale language model operates, which searches for information indicating a file storage location based on the similarity of a language vector of a natural language, a message sending means for sending a query message in a natural language to the large-scale language model to inquire about the storage location of the file; and a message receiving means for receiving from the large-scale language model a response message conveying information indicating the storage location of the file searched for by the large-scale language model using the language vector of the natural language of the query message.

2. 2. The information processing apparatus according to claim 1, wherein the message sending means and the message receiving means are text chats performed with the large-scale language model.

3. 3. The information processing device according to claim 1, further comprising a transmission means for transmitting information indicating a storage location of the file to the large-scale language model when an operation related to the file is performed on a device connected to the network.

4. 4. The information processing apparatus according to claim 3, wherein said transmitting means transmits information indicating a storage location of said file to said large-scale language model only when said file contains natural language.

5. the transmitting means transmits a user ID to the large-scale language model together with information indicating a storage location of the file; the message sending means sends the user ID together with the question message to the large-scale language model; The information processing device according to claim 3, characterized in that the message receiving means receives the response message when a user ID sent to the large-scale language model together with information indicating the storage location of the file searched by the large-scale language model using a language vector of the natural language of the question message matches a user ID sent to the large-scale language model together with the question message.

6. The information processing device according to claim 5, characterized in that the message receiving means receives an absence message indicating that the file does not exist instead of the response message when a user ID sent to the large-scale language model together with information indicating the storage location of the file searched by the large-scale language model using a language vector of the natural language of the question message does not match a user ID sent to the large-scale language model together with the question message.

7. The information processing device according to claim 3, characterized in that, when the operation related to the file is a new creation operation, the sending means sends the metadata of the file to the large-scale language model together with information indicating the storage location of the file, in order to have the large-scale language model perform a search using a language vector of the metadata of the file.

8. 8. The information processing apparatus according to claim 7, wherein the metadata of the file is at least one of the creation date and time of the file, the name of the creator, and the cause of creation.

9. 2. The information processing apparatus according to claim 1, wherein the information indicating the storage location of the file is a URL (Uniform Resource Locator).

10. 10. The information processing apparatus according to claim 9, wherein the communication protocol indicated by the scheme of the URL is HTTP (Hyper Text Transfer Protocol).

11. 10. The information processing apparatus according to claim 9, wherein the communication protocol indicated by the scheme of the URL is SMB (Server Message Block).

12. 2. The information processing apparatus according to claim 1, wherein the information processing apparatus is a PC (Personal Computer).

13. 2. The information processing apparatus according to claim 1, wherein the information processing apparatus is an image forming apparatus.

14. 2. The information processing apparatus according to claim 1, wherein the file is a document file.

15. 2. The information processing apparatus according to claim 1, wherein the file is an image file containing natural language.

16. A control method for an information processing device that can be connected via a network to a server on which a large-scale language model operates, which searches for information indicating a file storage location based on a similarity between language vectors of natural languages, the method comprising: a message sending step of sending a query message in a natural language to the large-scale language model to inquire about the storage location of the file; a message receiving step of receiving from the large-scale language model a response message conveying information indicating the storage location of the file searched for by the large-scale language model using a language vector of the natural language of the query message.

17. 2. A program for causing a computer to execute each means of the information processing apparatus according to claim 1.

18. An information processing system in which a server on which a large-scale language model runs and an information processing device can be connected via a network, The large-scale language model is an acquisition means for acquiring the file based on information indicating the storage location of the file; a storage means for storing in a database a data element that combines a language vector of the natural language included in the file and information indicating the storage location of the file; a search means for searching the database for information indicating a storage location of the file based on a similarity between a language vector of a natural language included in a query message from the information processing device and a language vector of a natural language included in the file; a response sending means for sending a response message to the information processing device to convey information indicating the storage location of the file searched for by the search means, The information processing device includes: a message sending means for sending a message inquiring about the storage location of the file in a natural language to the large-scale language model as the query message; and a message receiving means for receiving the response message from the large-scale language model.

19. 19. The information processing system according to claim 18, wherein the message sending means and the message receiving means are text chats between the information processing device and the large-scale language model.

20. 20. The information processing system according to claim 18, wherein the information processing device comprises a transmitting means for transmitting information indicating a storage location of the file to the large-scale language model when an operation related to the file is performed on a device connected to the network.

21. In the information processing device, the transmitting means transmits a user ID to the large-scale language model together with information indicating a storage location of the file; the message sending means sends the user ID together with the question message to the large-scale language model; In the large-scale language model, the storage means includes, in the database, the user ID transmitted from the information processing device to the large-scale language model together with the information indicating the storage location of the file, in a data element of the information indicating the storage location of the file; The information processing system according to claim 20, characterized in that the response sending means sends the response message to the information processing device when a user ID included in a data element of information indicating the storage location of the file searched by the search means using a language vector of the natural language of the question message matches a user ID sent from the information processing device to the large-scale language model together with the question message.

22. In the information processing device, the transmitting means transmits metadata of the file together with information indicating a storage location of the file to the large-scale language model when the operation related to the file is a new file creation operation; In the large-scale language model, the storage means includes, in the database, a language vector of the metadata of the file in a data element of information indicating a storage location of the file; The information processing system according to claim 20, characterized in that the search means searches for information indicating the storage location of the file based on the similarity between a language vector of the natural language contained in the question message from the information processing device and a language vector contained in a data element of the database.

23. The large-scale language model is a determination means for determining whether the file exists in the storage location of the file or in the vicinity of the storage location of the file based on information indicating the storage location of the file; 19. The information processing system according to claim 18, further comprising: a deletion means for deleting a data element including information indicating the storage location of the file from the database when the file does not exist in the storage location of the file or in the vicinity of the storage location of the file.

24. The information processing system of claim 23, wherein in the large-scale language model, the deletion means deletes from the database a data element containing information indicating the storage location of the file when the language vector for the file located at or near the storage location of the file differs from the language vector combined with information indicating the storage location of the file in a data element of the database.

25. The information processing system of claim 23, characterized in that the large-scale language model comprises a replacement means that, when a language vector for the file located near the storage location of the file matches a language vector combined with information indicating the storage location of the file in a data element of the database, replaces the information indicating the storage location of the file in the data element of the database containing information indicating the storage location of the file with information indicating the vicinity of the storage location of the file.

Images