Non-transitory computer-readable storage medium, information processing method

Abstract

To facilitate generation of a useful knowledge graph.SOLUTION: The information processing apparatus 100 identifies, from among the headings included in the object data 110, the heading corresponding to the first information 111 related to the text included in the object data 110. The information processing apparatus 100 generates the first knowledge graph 121 based on the first information 111 and the headings corresponding to the first information 111. The information processing apparatus 100 identifies, from among headings included in the object data 110, a heading corresponding to the second information 112 related to a graphical element included in the object data 110. The information processing apparatus 100 generates the second knowledge graph 122 related to the object data 110 based on the second information 112, the headings corresponding to the second information 112, and the generated first knowledge graph 121.SELECTED DRAWING: Figure 1

Description

[Technical Field]

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

[0002] 2. Description of the Related Art There is a conventional technique for generating a knowledge graph that includes a plurality of nodes each representing a different entity, and that represents the relationship between the entities corresponding to the nodes by edges connecting the nodes.

[0003] As a prior art, for example, there is a technique that integrates a large-scale language model and a knowledge graph and uses the knowledge graph for pre-training the large-scale language model. [Prior art documents] [Non-patent literature]

[0004] [Non-Patent Document 1] Pan, Shirui, et al. “Unifying large language models and knowledge graphs: A roadmap.” IEEE Transactions on Knowledge and Data Engineering (2024). Summary of the Invention [Problem to be solved by the invention]

[0005] However, conventional techniques sometimes have difficulty generating useful knowledge graphs. For example, there may be cases where target data representing design documents or development regulations is divided into multiple fragments, and an overall knowledge graph is generated by collecting partial knowledge graphs based on each fragment. In this case, it is not possible to generate a knowledge graph that represents the relationships between entities that appear in different fragments.

[0006] In one aspect, the present invention aims to make it easier to generate useful knowledge graphs. [Means for solving the problem]

[0007] According to one embodiment, an information processing program, an information processing method, and an information processing device are proposed that acquire first information about sentences included in target data, second information about graphic elements included in the target data, and third information about headings included in the target data, and by referring to the acquired third information, generate a first knowledge graph about the target data based on either the acquired first information or the acquired second information, and a heading among the headings included in the target data that corresponds to the one piece of information, and by referring to the acquired third information, generate a second knowledge graph about the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other piece of information, and the generated first knowledge graph. [Effects of the Invention]

[0008] According to one aspect, it becomes possible to facilitate the generation of useful knowledge graphs. [Brief explanation of the drawings]

[0009] [Figure 1] FIG. 1 is an explanatory diagram illustrating an example of an information processing method according to an embodiment. [Figure 2] FIG. 2 is an explanatory diagram illustrating an example of an information processing system 200. As shown in FIG. [Figure 3] FIG. 3 is a block diagram showing an example of the hardware configuration of the information processing device 100. As shown in FIG. [Figure 4] FIG. 4 is a block diagram showing an example of the functional configuration of the information processing device 100. As shown in FIG. [Figure 5] FIG. 5 is an explanatory diagram (part 1) showing an example of the operation of the information processing device 100. [Figure 6]FIG. 6 is an explanatory diagram (part 2) showing an example of the operation of the information processing device 100. [Figure 7] FIG. 7 is an explanatory diagram (part 3) showing an example of the operation of the information processing device 100. [Figure 8] FIG. 8 is an explanatory diagram (part 4) showing an example of the operation of the information processing device 100. [Figure 9] FIG. 9 is an explanatory diagram (part 5) showing an example of the operation of the information processing device 100. [Figure 10] FIG. 10 is an explanatory diagram (part 6) showing an example of the operation of the information processing device 100. [Figure 11] FIG. 11 is an explanatory diagram showing a specific example of the operation of the information processing device 100. [Figure 12] FIG. 12 is an explanatory diagram (part 1) showing an example of document data 1100. [Figure 13] FIG. 13 is an explanatory diagram (part 2) showing an example of the document data 1100. [Figure 14] FIG. 14 is an explanatory diagram (part 1) showing an example of a Markdown document 1110. [Figure 15] FIG. 15 is an explanatory diagram (part 2) showing an example of a Markdown document 1110. [Figure 16] FIG. 16 is an explanatory diagram showing an example of the headline-attached sentence fragment data 1141. As shown in FIG. [Figure 17] FIG. 17 is an explanatory diagram showing an example of the captioned table fragment data 1142. [Figure 18] FIG. 18 is an explanatory diagram (part 1) showing an example of the effect. [Figure 19] FIG. 19 is an explanatory diagram (part 2) showing an example of the effect. [Figure 20] FIG. 20 is an explanatory diagram (part 3) showing an example of the effect. [Figure 21] FIG. 21 is a flowchart illustrating an example of the overall processing procedure. DETAILED DESCRIPTION OF THE INVENTION

[0010] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, an information processing program, an information processing method, and an information processing device according to embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[0011] (An example of an information processing method according to an embodiment) 1 is an explanatory diagram illustrating an example of an information processing method according to an embodiment. The information processing device 100 is a computer that facilitates generating a knowledge graph. The information processing device 100 is, for example, a server or a PC (Personal Computer).

[0012] A knowledge graph includes multiple nodes, each representing a different entity. A knowledge graph includes multiple edges connecting the different nodes. A knowledge graph uses edges connecting nodes to represent relationships between the entities corresponding to the nodes. An entity is also called, for example, a substance. An entity is, for example, a sentence.

[0013] Conventionally, for example, when creating test specifications, it is sometimes desirable to generate and utilize a knowledge graph. Specifically, when creating test specifications manually, test items are likely to be overlooked, and it is difficult to grasp the latest test items in response to system changes or function updates, so it is desirable to generate and utilize a knowledge graph.

[0014] There are methods for automatically generating knowledge graphs. For example, one method involves analyzing target data and generating a knowledge graph that represents the relationships between at least some of the entities that appear (contained) in the target data. However, with this method, the generated knowledge graph may not represent the relationships between different entities that appear relatively close to each other in the target data. As a result, this method may not be able to generate a useful knowledge graph.

[0015] In contrast to this, a conceivable approach is to divide the target data into multiple fragments and then generate an overall knowledge graph by collecting partial knowledge graphs based on each fragment. The target data may represent design documents or development regulations, for example. With this approach, by dividing the target data into multiple fragments and then generating an overall knowledge graph, it is expected that the generated knowledge graph will represent the relationships between different entities that appear in the same fragment. Therefore, with this approach, it is expected that the generated knowledge graph will represent the relationships between different entities that appear relatively close to each other in the target data.

[0016] However, this method may have difficulty in generating a useful knowledge graph. For example, this method divides the target data into multiple fragments, making it impossible to generate a knowledge graph that represents the relationships between different entities that appear in different fragments.

[0017] Therefore, in this embodiment, an information processing method that can make it easier to generate a useful knowledge graph will be described.

[0018] In FIG. 1, target data 110 exists. The target data 110 includes, for example, one or more sentences. The target data 110 includes, for example, one or more diagrams. The target data 110 includes, for example, one or more headings. The one or more headings have, for example, a hierarchical structure. For example, a sentence appears in an area of ​​the target data 110 that corresponds to a hierarchy represented by a combination of one or more headings. For example, a diagram appears in an area of ​​the target data 110 that corresponds to a hierarchy represented by a combination of one or more headings.

[0019] (1-1) The information processing device 100 acquires one or more pieces of first information 111, one or more pieces of second information 112, and one or more pieces of third information 113. The one or more pieces of first information 111 include one or more pieces of information relating to different sentences that appear (are included) in the target data 110. The one or more pieces of first information 111 include, for example, one or more pieces of information relating to different sentences among a plurality of sentences, each of which has a predetermined length, that have been divided from the target data. The predetermined length is, for example, one sentence or one paragraph. The one or more pieces of first information 111 specifically include one or more pieces of fragment data that each represent a different sentence.

[0020] The one or more pieces of second information 112 include one or more pieces of information about different graphical elements that appear (are included) in the target data 110. A graphical element is, for example, the entirety of a diagram or table. A graphical element may be, for example, a part that forms a diagram or table. A graphical element is, specifically, a record that forms a table. The one or more pieces of second information 112 include, for example, one or more pieces of information about different records among a plurality of records that form tables separated from the target data. The one or more pieces of second information 112 specifically include one or more pieces of data that represent different records. The one or more pieces of second information 112 may include, for example, one or more pieces of information about different entities among a plurality of entities that form diagrams separated from the target data. The diagrams are graphs, flowcharts, or the like. The second information 112 may be generated, for example, by performing image analysis on the diagram and identifying entities depicted in the diagram.

[0021] The one or more pieces of third information 113 include one or more pieces of information relating to different headings that appear (are included) in the target data 110. The information processing device 100 may acquire the one or more pieces of first information 111, one or more pieces of second information 112, and one or more pieces of third information 113 by generating them based on the target data 110. Note that the information processing device 100 may acquire the first information 111, the second information 112, and the third information 113 after acquiring the target data 110. Alternatively, the information processing device 100 may acquire the first information 111, the second information 112, and the third information 113 without acquiring the target data 110.

[0022] (1-2) The information processing device 100 selects one or more pieces of acquired first information 111 and one or more pieces of acquired second information 112. Preferably, the information processing device 100 selects, for example, one or more pieces of first information 111. The information processing device 100 refers to the one or more pieces of acquired third information 113 and identifies, from the one or more headlines, a headline that corresponds to each of the selected pieces of information. The information processing device 100 generates a first knowledge graph 121 related to the target data 110 based on the selected piece of information and the identified headline that corresponds to each of the selected pieces of information.

[0023] This allows the information processing device 100 to generate the first knowledge graph 121 so as to appropriately represent the relationship between different entities that appear in the same first information 111 or the same second information 112. Furthermore, the information processing device 100 can generate the first knowledge graph 121 so as to represent the relationship between different entities that appear in different first information 111 or different second information 112 via headings.

[0024] (1-3) The information processing device 100 selects the other of the one or more pieces of first information 111 that have been acquired and the one or more pieces of second information 112 that have been acquired. Preferably, the information processing device 100 selects, for example, one or more pieces of second information 112. The information processing device 100 refers to the one or more pieces of third information 113 that have been acquired and identifies, from the one or more headlines, a headline that corresponds to each of the selected pieces of other information. The information processing device 100 generates a second knowledge graph 122 related to the target data 110 based on the selected pieces of other information, the identified headline that corresponds to each of the selected pieces of other information, and the generated first knowledge graph 121.

[0025] This allows the information processing device 100 to generate the second knowledge graph 122 so as to appropriately represent the relationships between different entities that appear in the same first information 111 or the same second information 112. Furthermore, the information processing device 100 can generate the second knowledge graph 122 so as to represent the relationships between different entities that appear in different first information 111 or different second information 112 via headings.

[0026] Furthermore, the information processing device 100 can determine which relationships between different entities the second knowledge graph 122 should preferably represent, taking into account the relationships between different entities represented by the first knowledge graph 121. In this way, the information processing device 100 can generate a useful second knowledge graph 122. "Useful" means that the second knowledge graph 122 appropriately represents relationships between entities that are preferable to present to the user.

[0027] Here, a case has been described in which the information processing device 100 selects one or more pieces of first information 111 when selecting one or more pieces of acquired first information 111 and one or more pieces of acquired second information 112, but this is not limiting. For example, there may be a case in which the information processing device 100 selects one or more pieces of second information 112 when selecting one or more pieces of acquired first information 111 and one or more pieces of acquired second information 112.

[0028] Here, the case where the functions of the information processing device 100 are realized by a single computer has been described, but this is not limiting. For example, the functions of the information processing device 100 may be realized by cooperation of multiple computers. For example, the functions of the information processing device 100 may be realized on the cloud.

[0029] (An example of the information processing system 200) Next, an example of an information processing system 200 to which the information processing device 100 shown in FIG. 1 is applied will be described with reference to FIG.

[0030] 2 is an explanatory diagram showing an example of an information processing system 200. In FIG. 2, the information processing system 200 includes an information processing device 100, an information storage device 201, a generation processing device 202, and a client device 203.

[0031] In the information processing system 200, the information processing device 100 and the information storage device 201 are connected via a wired or wireless network 210. The network 210 is, for example, a local area network (LAN), a wide area network (WAN), or the Internet. In the information processing system 200, the information processing device 100 and the generation processing device 202 are connected via the wired or wireless network 210. In the information processing system 200, the information processing device 100 and the client device 203 are connected via the wired or wireless network 210.

[0032] The information processing device 100 is a computer that facilitates the generation of a useful knowledge graph. The information processing device 100 receives, for example, a generation request from a client device 203 requesting the generation of a knowledge graph related to target data. The generation request includes, for example, target data. The target data includes, for example, one or more sentences, one or more tables, and one or more headings. The one or more headings have, for example, a hierarchical structure. The information processing device 100 acquires the target data based on the generation request.

[0033] The information processing device 100, for example, identifies one or more headings based on the target data, and generates one or more sentence fragment data each representing a different sentence, and one or more table fragment data each representing a different record. A record corresponds to one row forming a table. The information processing device 100 assigns a heading to each of the sentence fragment data of the one or more sentence fragment data based on the identified one or more headings. The information processing device 100 assigns a heading to each of the table fragment data of the one or more table fragment data based on the identified one or more headings.

[0034] The information processing device 100 generates a first prompt requesting the generation of a first knowledge graph based on one or more pieces of sentence fragment data, and transmits the first prompt to the generation processing device 202. The information processing device 100 receives the first knowledge graph from the generation processing device 202. The information processing device 100 transmits the first knowledge graph to the information storage device 201.

[0035] The information processing device 100 generates a second prompt requesting generation of a second knowledge graph based on one or more table fragment data and the received first knowledge graph, and transmits the second prompt to the generation processing device 202. The information processing device 100 receives the second knowledge graph from the generation processing device 202. The information processing device 100 transmits the second knowledge graph to the information storage device 201. The information processing device 100 transmits the received second knowledge graph to the client device 203. The information processing device 100 is, for example, a server or a PC.

[0036] The information storage device 201 is a computer that stores a knowledge graph. The information storage device 201 receives the knowledge graph from the information processing device 100. The information storage device 201 stores the received knowledge graph. The information storage device 201 is, for example, a server or a PC.

[0037] The generation processing device 202 is, for example, a computer having a large-scale language model. The large-scale language model has a function that enables generation of a knowledge graph in response to, for example, a prompt. Specifically, the large-scale language model has a function that generates input data for a generation unit that generates a knowledge graph in response to a prompt. Specifically, the large-scale language model may have a function that generates a knowledge graph in response to a prompt.

[0038] The generation processing device 202 receives, for example, a first prompt from the information processing device 100. The generation processing device 202 generates a first knowledge graph by inputting the received first prompt into a large-scale language model. The generation processing device 202 transmits the generated first knowledge graph to the information processing device 100. The generation processing device 202 receives, for example, a second prompt from the information processing device 100. The generation processing device 202 generates a second knowledge graph by inputting the received second prompt into a large-scale language model. The generation processing device 202 transmits the generated second knowledge graph to the information processing device 100. The generation processing device 202 is, for example, a server or a PC.

[0039] The client device 203 is a computer used by a worker who wishes to use a knowledge graph. The client device 203 transmits a generation request to the information processing device 100, requesting that a knowledge graph be generated, based on, for example, an operational input by the worker. The client device 203 receives, for example, a knowledge graph from the information processing device 100. The client device 203 outputs, for example, the received knowledge graph so that the worker can refer to it. The client device 203 is, for example, a PC, a tablet terminal, or a smartphone.

[0040] Here, the case where the information processing device 100 is a device different from the information storage device 201 has been described, but this is not limiting. For example, the information processing device 100 may have the function of the information storage device 201 and may also operate as the information storage device 201. In this case, the information processing system 200 may not include the information storage device 201.

[0041] Here, the case where the information processing device 100 is a device different from the generation processing device 202 has been described, but this is not limiting. For example, the information processing device 100 may have the function of the generation processing device 202 and may also operate as the generation processing device 202. In this case, the information processing system 200 may not include the generation processing device 202.

[0042] Here, the case where the information processing device 100 is a device different from the client device 203 has been described, but this is not limiting. For example, the information processing device 100 may have the function of the client device 203 and may also operate as the client device 203. In this case, the information processing system 200 may not include the client device 203.

[0043] (Example of hardware configuration of information processing device 100) Next, an example of the hardware configuration of the information processing device 100 will be described with reference to FIG.

[0044] Fig. 3 is a block diagram showing an example of the hardware configuration of the information processing device 100. In Fig. 3, the information processing device 100 has a CPU (Central Processing Unit) 301, a memory 302, and a network I / F (Interface) 303. The information processing device 100 also has a recording medium I / F 304, a recording medium 305, a display 306, and an input device 307. The components are connected to each other via a bus 300.

[0045] Here, CPU 301 is responsible for overall control of information processing device 100. Memory 302 includes, for example, a read-only memory (ROM), a random access memory (RAM), and a flash ROM. Specifically, for example, the flash ROM or ROM stores various programs, and RAM is used as a work area for CPU 301. The programs stored in memory 302 are loaded into CPU 301, causing CPU 301 to execute coded processes.

[0046] The network I / F 303 is connected to the network 210 via a communication line, and is connected to other computers via the network 210. The network I / F 303 manages the internal interface with the network 210 and controls the input and output of data from other computers. The network I / F 303 is, for example, a modem or a LAN adapter.

[0047] The recording medium I / F 304 controls reading and writing of data from and to the recording medium 305 under the control of the CPU 301. The recording medium I / F 304 is, for example, a disk drive, a solid state drive (SSD), or a universal serial bus (USB) port. The recording medium 305 is a non-volatile memory that stores data written under the control of the recording medium I / F 304. The recording medium 305 is, for example, a disk, a semiconductor memory, or a USB memory. The recording medium 305 may be detachable from the information processing device 100.

[0048] The display 306 displays data such as a cursor, an icon, a toolbox, a document, an image, or function information. The display 306 is, for example, a CRT (Cathode Ray Tube), a liquid crystal display, or an organic EL (Electroluminescence) display. The input device 307 has keys for inputting characters, numbers, various instructions, etc., and inputs data. The input device 307 is, for example, a keyboard or a mouse. The input device 307 may also be, for example, a touch panel input pad or a numeric keypad.

[0049] The information processing device 100 may have, in addition to the above-described components, for example, a camera. The information processing device 100 may have, in addition to the above-described components, for example, a printer, a scanner, a microphone, or a speaker. The information processing device 100 may have, for example, a plurality of recording medium I / Fs 304 and recording media 305. The information processing device 100 may not have, for example, a display 306 or an input device 307. The information processing device 100 may not have, for example, a recording medium I / F 304 or a recording medium 305.

[0050] (Example of hardware configuration of information storage device 201) An example of the hardware configuration of the information storage device 201 is similar to the example of the hardware configuration of the information processing device 100 shown in FIG. 3, and therefore a description thereof will be omitted.

[0051] (Example of hardware configuration of generation processing device 202) An example of the hardware configuration of the generation processing device 202 is similar to the example of the hardware configuration of the information processing device 100 shown in FIG. 3, and therefore a description thereof will be omitted.

[0052] (Example of hardware configuration of client device 203) An example of the hardware configuration of the client device 203 is similar to the example of the hardware configuration of the information processing device 100 shown in FIG. 3, and therefore a description thereof will be omitted.

[0053] (Example of functional configuration of information processing device 100) Next, an example of the functional configuration of the information processing device 100 will be described with reference to FIG.

[0054] 4 is a block diagram showing an example of the functional configuration of the information processing device 100. The information processing device 100 includes a storage unit 400, an acquisition unit 401, a division unit 402, a first generation unit 403, a second generation unit 404, and an output unit 405.

[0055] The storage unit 400 is realized by, for example, a storage area such as the memory 302 or the recording medium 305 shown in Fig. 3. In the following, a case where the storage unit 400 is included in the information processing device 100 will be described, but this is not limiting. For example, the storage unit 400 may be included in a device different from the information processing device 100, and the stored contents of the storage unit 400 may be accessible from the information processing device 100.

[0056] The acquiring unit 401 to the output unit 405 function as an example of a control unit. Specifically, the acquiring unit 401 to the output unit 405 realize their functions by causing the CPU 301 to execute a program stored in a storage area such as the memory 302 or the recording medium 305 shown in Fig. 3, or by using the network I / F 303. The processing results of each functional unit are stored in a storage area such as the memory 302 or the recording medium 305 shown in Fig. 3, for example.

[0057] The memory unit 400 stores various information that is referenced or updated during processing by each functional unit. The memory unit 400 stores, for example, target data. The target data includes, for example, one or more sentences, one or more diagrams, and one or more headings. The one or more headings have, for example, a hierarchical structure. For example, a sentence appears in an area of ​​the target data that corresponds to a hierarchy represented by any combination of one or more headings. For example, a diagram appears in an area of ​​the target data that corresponds to a hierarchy represented by any combination of one or more headings. Specifically, the target data represents design documents, development regulations, etc. The target data is acquired, for example, by the acquisition unit 401.

[0058] The storage unit 400 stores, for example, one or more pieces of first information relating to different sentences appearing (contained) in the target data. The one or more pieces of first information include, for example, one or more pieces of information relating to different sentences among a plurality of sentences, each of which has a predetermined length, divided from the target data. Specifically, the one or more pieces of first information include sentence fragment data representing different sentences. The first information is generated, for example, by the dividing unit 402. The first information may be acquired, for example, by the acquiring unit 401.

[0059] The storage unit 400 stores, for example, one or more pieces of second information regarding different diagram elements that appear (are included) in the target data. The diagram element is, for example, a diagram or a table. The diagram element may be, for example, a part that forms a diagram or a table. Specifically, the diagram element is, for example, a record that forms a table. The one or more pieces of second information include, for example, one or more pieces of information regarding different records among multiple records that form a table separated from the target data. Specifically, the one or more pieces of second information include table fragment data that represent different records. The one or more pieces of second information may include, for example, one or more pieces of information regarding different entities among multiple entities that form a diagram separated from the target data. The diagram is, for example, a graph or a flowchart. The entity is, for example, a word. The second information is generated by, for example, the dividing unit 402. The second information may be acquired by, for example, the acquiring unit 401.

[0060] The storage unit 400 stores, for example, one or more pieces of third information relating to different headings that appear (are included) in the target data. The third information is generated, for example, by the dividing unit 402. The third information may be acquired, for example, by the acquiring unit 401.

[0061] The storage unit 400 may store, for example, a language model that enables generation of a knowledge graph in response to an input prompt. The language model is a large-scale language model. The language model is, for example, preset by a user.

[0062] The acquisition unit 401 acquires various types of information used in processing by each functional unit. The acquisition unit 401 stores the acquired various types of information in the storage unit 400 or outputs it to each functional unit. The acquisition unit 401 may also output the various types of information stored in the storage unit 400 to each functional unit. The acquisition unit 401 acquires various types of information based on, for example, a user's operation input. The acquisition unit 401 may receive various types of information from, for example, a device different from the information processing device 100.

[0063] The acquiring unit 401 acquires a generation request requesting the generation of a knowledge graph related to target data. The generation request includes, for example, the target data. The generation request may include, for example, one or more pieces of first information, one or more pieces of second information, and one or more pieces of third information. The acquiring unit 401 acquires, for example, the generation request. Specifically, the acquiring unit 401 acquires the generation request by accepting an input of the generation request. Specifically, the acquiring unit 401 may acquire the generation request by receiving the generation request from another computer. The other computer is, for example, the client device 203.

[0064] The acquiring unit 401 acquires, for example, target data. Specifically, the acquiring unit 401 acquires the target data by accepting input of the target data. Specifically, the acquiring unit 401 may acquire the target data by receiving the target data from another computer. The other computer is, for example, the client device 203. Specifically, the acquiring unit 401 may acquire the target data by extracting the target data from the generation request.

[0065] The acquiring unit 401 acquires, for example, one or more pieces of first information. Specifically, the acquiring unit 401 acquires the one or more pieces of first information by accepting input of the one or more pieces of first information. Specifically, the acquiring unit 401 may acquire the one or more pieces of first information by receiving the one or more pieces of first information from another computer. The other computer is, for example, the client device 203. Specifically, the acquiring unit 401 may acquire the one or more pieces of first information by extracting the one or more pieces of first information from a generation request.

[0066] The acquiring unit 401 acquires, for example, one or more pieces of second information. Specifically, the acquiring unit 401 acquires the one or more pieces of second information by accepting input of the one or more pieces of second information. Specifically, the acquiring unit 401 may acquire the one or more pieces of second information by receiving the one or more pieces of second information from another computer. The other computer is, for example, the client device 203. Specifically, the acquiring unit 401 may acquire the one or more pieces of second information by extracting the one or more pieces of second information from a generation request.

[0067] The acquiring unit 401 acquires, for example, one or more pieces of third information. Specifically, the acquiring unit 401 acquires the one or more pieces of third information by accepting input of the one or more pieces of third information. Specifically, the acquiring unit 401 may acquire the one or more pieces of third information by receiving the one or more pieces of third information from another computer. The other computer is, for example, the client device 203. Specifically, the acquiring unit 401 may acquire the one or more pieces of third information by extracting the one or more pieces of third information from a generation request.

[0068] The acquisition unit 401 may receive a start trigger to start processing of any of the functional units. The start trigger may be, for example, a predetermined operation input by a user. The start trigger may be, for example, reception of predetermined information from another computer. The start trigger may be, for example, output of predetermined information by any of the functional units. The acquisition unit 401 receives, for example, acquisition of a generation request as a start trigger to start processing by the division unit 402, the first generation unit 403, and the second generation unit 404.

[0069] The dividing unit 402 obtains the first information by generating one or more pieces of first information. For example, the dividing unit 402 divides the target data into a plurality of sentences, each of which has a predetermined length. The predetermined length is, for example, one sentence or one paragraph. The dividing unit 402 generates, for example, one or more pieces of first information relating to different sentences from the divided plurality of sentences. In this way, the dividing unit 402 can obtain the first information, which is a fragment to be used when generating a knowledge graph.

[0070] The dividing unit 402 obtains the second information by generating one or more pieces of second information. For example, the dividing unit 402 divides a plurality of records, each of which forms one of the tables, from the target data. For example, the dividing unit 402 generates one or more pieces of second information related to different records from the divided plurality of records. For example, the dividing unit 402 may divide a plurality of entities, each of which forms one of the diagrams, from the target data, and generate one or more pieces of second information related to different entities from the divided plurality of entities. In this way, the dividing unit 402 can obtain the second information, which is a fragment used when generating a knowledge graph.

[0071] The dividing unit 402 obtains the third information by generating one or more pieces of third information. For example, the dividing unit 402 identifies one or more headings that appear in the target data. For example, the dividing unit 402 generates one or more pieces of third information related to different headings from the identified one or more headings. The one or more pieces of third information may be a bundle of information. This allows the dividing unit 402 to identify one or more headings to be used when generating a knowledge graph.

[0072] The first generation unit 403 generates a first knowledge graph related to the target data. The first generation unit 403 selects, for example, one or more pieces of first information acquired by the acquisition unit 401 or the division unit 402, and one or more pieces of second information acquired by the acquisition unit 401 or the division unit 402. Specifically, it is preferable that the first generation unit 403 selects one or more pieces of first information.

[0073] The first generating unit 403, for example, refers to one or more pieces of third information acquired by the acquiring unit 401 or the dividing unit 402, and identifies a heading corresponding to each of the selected pieces of information from among one or more headings appearing in the target data. Specifically, the first generating unit 403 identifies a first heading corresponding to each piece of first information, or a second heading corresponding to each piece of second information. The first heading is each of the one or more headings that form a combination of headings indicating an area to which the first information belongs in the target data. The second heading is each of the one or more headings that form a combination of headings indicating an area to which the second information belongs in the target data.

[0074] The first generation unit 403 generates a first knowledge graph based on, for example, one of the selected items and the identified headings corresponding to each piece of information in the selected item. Specifically, the first generation unit 403 assigns the identified headings corresponding to each piece of information in the selected item to the information. The first generation unit 403 generates the first knowledge graph based on each piece of information in the selected item to which the headings are assigned.

[0075] Specifically, there may be a case where the first generation unit 403 selects one or more pieces of first information. In this case, the first generation unit 403 specifically assigns, to each piece of first information among the one or more selected pieces of first information, a specified first heading that corresponds to the first information, from among one or more headings. The first generation unit 403 specifically generates a first knowledge graph based on the one or more pieces of first information to which the first headings have been assigned.

[0076] More specifically, the first generation unit 403 generates a prompt that includes one or more pieces of first information, each of which is assigned a first heading, and requests that a first knowledge graph be generated based on the respective pieces of first information. More specifically, the first generation unit 403 generates the first knowledge graph by inputting the generated prompt to a language model. More specifically, the first generation unit 403 may generate and acquire the first knowledge graph by transmitting the generated prompt to the generation processing device 202.

[0077] Specifically, there may be a case where the first generation unit 403 selects one or more pieces of second information. In this case, the first generation unit 403 specifically assigns, to each piece of second information among the one or more selected pieces of second information, a specified second heading that corresponds to the second information, from among one or more headings. The first generation unit 403 specifically generates a first knowledge graph based on the one or more pieces of second information to which the second headings have been assigned.

[0078] More specifically, the first generation unit 403 generates a prompt that includes one or more pieces of second information, each of which is assigned a second heading, and requests that a first knowledge graph be generated based on the respective pieces of second information. More specifically, the first generation unit 403 generates the first knowledge graph by inputting the generated prompt to a language model. More specifically, the first generation unit 403 may generate and acquire the first knowledge graph by transmitting the generated prompt to the generation processing device 202.

[0079] This allows the first generation unit 403 to generate a first knowledge graph to appropriately represent the relationships between different entities that appear in the same first information or the same second information. Furthermore, the first generation unit 403 can generate a first knowledge graph to represent, via headings, the relationships between different entities that appear in different first information or different second information.

[0080] Furthermore, when the first generation unit 403 selects, for example, one or more pieces of first information, it can generate a first knowledge graph that takes into consideration the context of sentence portions that appear in the target data. Therefore, when the first generation unit 403 selects, for example, one or more pieces of first information, it can easily analyze the relationships between different entities and can easily generate a first knowledge graph that accurately represents the relationships between different entities.

[0081] The second generation unit 404 generates a second knowledge graph related to the target data. For example, the second generation unit 404 selects one of the one or more pieces of first information acquired by the acquisition unit 401 or the division unit 402 and the one or more pieces of second information acquired by the acquisition unit 401 or the division unit 402, the other being different from the one or more pieces of second information selected by the first generation unit 403. Specifically, the second generation unit 404 preferably selects one or more pieces of second information.

[0082] The second generating unit 404, for example, refers to one or more pieces of third information acquired by the acquiring unit 401 or the dividing unit 402, and identifies, from among the one or more headings, a heading that corresponds to each of the selected other pieces of information. Specifically, the second generating unit 404 identifies a first heading that corresponds to each piece of first information, or a second heading that corresponds to each piece of second information. The first heading is, from among the one or more headings, each heading that forms a combination of headings that indicates an area to which the first information belongs in the target data. The second heading is, from among the one or more headings, each heading that forms a combination of headings that indicates an area to which the second information belongs in the target data.

[0083] The second generation unit 404 generates a second knowledge graph based on, for example, the selected one of the others, the identified headings corresponding to each piece of information about the selected one of the others, and the first knowledge graph generated by the first generation unit 403. Specifically, the second generation unit 404 assigns the identified headings corresponding to each piece of information about the selected one of the others to that information. The second generation unit 404 generates the second knowledge graph based on the information about the one of the others to which headings have been assigned, and the first knowledge graph generated by the first generation unit 403.

[0084] Specifically, there may be a case where the second generation unit 404 selects one or more pieces of first information. In this case, the second generation unit 404 specifically assigns, to each piece of first information among the one or more selected pieces of first information, a specified first heading that corresponds to the first information, from among one or more headings. The second generation unit 404 specifically generates a second knowledge graph based on the one or more pieces of first information to which the first headings have been assigned, and the first knowledge graph.

[0085] More specifically, the second generation unit 404 generates a prompt that requests the language model to generate a second knowledge graph based on the first knowledge graph and one or more pieces of first information, each of which includes a first heading. More specifically, the second generation unit 404 generates the second knowledge graph by inputting the generated prompt to the language model. More specifically, the second generation unit 404 may generate and acquire the second knowledge graph by transmitting the generated prompt to the generation processing device 202.

[0086] Specifically, there may be a case where the second generation unit 404 selects one or more pieces of second information. In this case, the second generation unit 404 specifically assigns, to each piece of second information among the one or more selected pieces of second information, a specified second heading that corresponds to the second information, from among one or more headings. The second generation unit 404 specifically generates a second knowledge graph based on the one or more pieces of second information to which the second headings have been assigned, and the first knowledge graph.

[0087] More specifically, the second generation unit 404 generates a prompt for the language model, which includes one or more pieces of second information, each of which is assigned a second heading, and requests that a second knowledge graph be generated based on the respective pieces of second information and the first knowledge graph. More specifically, the second generation unit 404 generates the second knowledge graph by inputting the generated prompt to the language model. More specifically, the second generation unit 404 may generate and acquire the second knowledge graph by transmitting the generated prompt to the generation processing device 202.

[0088] This allows the second generation unit 404 to generate a second knowledge graph to appropriately represent relationships between different entities that appear in the same first information or the same second information. Furthermore, the second generation unit 404 can generate a second knowledge graph to represent relationships between different entities that appear in different first information or different second information via headings. Furthermore, the second generation unit 404 can determine which relationships between different entities the second knowledge graph should preferably represent, taking into account the relationships between different entities represented in the first knowledge graph.

[0089] For example, when the first generation unit 403 selects one or more pieces of first information, the second generation unit 404 can use the first knowledge graph to generate a second knowledge graph that takes into account the context of sentence portions that appear in the target data. Therefore, when the first generation unit 403 selects one or more pieces of first information, the information processing device 100 can easily analyze the relationships between different entities and can easily generate a second knowledge graph that accurately represents the relationships between different entities. In this way, the second generation unit 404 can generate a useful second knowledge graph.

[0090] The output unit 405 outputs the processing result of at least one of the functional units. The output format is, for example, display on a display, printout to a printer, transmission to an external device via the network I / F 303, or storage in a storage area such as the memory 302 or the recording medium 305. In this way, the output unit 405 can notify the user of the processing result of at least one of the functional units, thereby improving the convenience of the information processing device 100.

[0091] The output unit 405 outputs, for example, the second knowledge graph generated by the second generation unit 404. Specifically, the output unit 405 outputs the second knowledge graph generated by the second generation unit 404 so that it can be referenced by a user. Specifically, the output unit 405 may transmit the second knowledge graph generated by the second generation unit 404 to another computer. In this way, the output unit 405 can make the useful second knowledge graph available externally.

[0092] Here, the case where the information processing device 100 includes the dividing unit 402, the first generating unit 403, and the second generating unit 404 has been described, but this is not limiting. For example, the information processing device 100 may not include any of the functions. Specifically, the information processing device 100 may not include the dividing unit 402. In this case, specifically, the information processing device 100 uses another computer that includes the dividing unit 402.

[0093] (An example of the operation of the information processing device 100) Next, an example of the operation of the information processing device 100 will be described with reference to FIGS.

[0094] 5 to 10 are explanatory diagrams showing an example of the operation of the information processing device 100. In FIG. 5, the information processing device 100 acquires target data 500. The target data 500 is, for example, a Markdown document corresponding to a document such as a design document or development regulations. The target data 500 includes, for example, multiple headings having a hierarchical structure. The target data 500 includes one or more sentences. The target data 500 includes one or more tables.

[0095] Each of the one or more sentences exists in an area within the target data 500 that corresponds to a hierarchical level represented by a combination of one or more of the multiple headings. In the example of Figure 5, the sentence "-zzzzzzyyyy -wwwwwvvvv" exists in an area within the target data 500 that corresponds to a hierarchical level represented by a combination of the heading "###1.AAAIIII" and its lower-level heading "####2.BBBJJJJ."

[0096] Each of the one or more tables exists in an area within the target data 500 that corresponds to a hierarchical level represented by a combination of one or more of the multiple headings. In the example of Figure 5, the table relating to the items "EEE" and "FFF" exists in an area within the target data 500 that corresponds to a hierarchical level represented by the combination of the heading "###1.AAAIIII" and its lower-level heading "####2.BBBJJJJ."

[0097] (5-1) The information processing device 100 performs a fragmentation process on the target data 500. Through the fragmentation process, the information processing device 100 generates, for example, header data 503 that groups together multiple headers. In the example of Fig. 5, the header data 503 specifically includes the header "###1.AAAIIII" and the header "####2.BBBJJJJ".

[0098] Through the fragmentation process, the information processing device 100 divides, for example, one or more sentences, each of a predetermined length, from the target data 500, and generates one or more pieces of sentence fragment data 501, each representing a different sentence from the one or more divided sentences. In the example of Fig. 5, the one or more pieces of sentence fragment data 501 specifically include sentence fragment data 501 representing "-xxxxxxmmm -yyyyyynnnn".

[0099] Through fragmentation processing, the information processing device 100, for example, divides one or more records from the target data 500, each of which forms one of the tables, and generates one or more table fragment data 502, each of which represents a different record from the one or more divided records.

[0100] (5-2) The information processing device 100 refers to the target data 500 and the heading data 503, and performs heading assignment processing on the one or more pieces of generated sentence fragment data 501 and the one or more pieces of generated table fragment data 502.

[0101] For example, the information processing device 100 performs a heading assignment process to identify, for each of the one or more pieces of sentence fragment data 501, one or more headings corresponding to the sentence represented by the sentence fragment data 501. Specifically, the information processing device 100 identifies, from among the multiple headings, each heading that forms a combination of one or more headings that indicates an area to which the sentence represented by the sentence fragment data 501 in the target data 500 belongs.

[0102] For example, the information processing device 100 performs a heading assignment process to assign one or more identified headings to each piece of sentence fragment data 501, thereby generating headline-attached sentence fragment data 511 corresponding to the sentence fragment data 501. In this way, the information processing device 100 can acquire one or more pieces of headline-attached sentence fragment data 511 that are the basis for generating a knowledge graph.

[0103] For example, the information processing device 100 performs a heading assignment process to identify, for each of the one or more table fragment data 502, one or more headings corresponding to the record represented by the table fragment data 502. Specifically, the information processing device 100 identifies, from among the multiple headings, each heading that forms a combination of one or more headings that indicates an area to which the record represented by the table fragment data 502 in the target data 500 belongs.

[0104] The information processing device 100 generates captioned table fragment data 512 corresponding to each table fragment data 502 by, for example, performing a captioning process to assign one or more identified captions to the respective table fragment data 502. This allows the information processing device 100 to acquire one or more captioned table fragment data 512 that are the basis for generating a knowledge graph. Next, we move on to the description of FIG. 6.

[0105] 6, (6-1) the information processing device 100 generates a text knowledge graph 600 that represents relationships between entities that appear in a text portion, based on one or more headline-attached text fragment data 511. The information processing device 100 generates first knowledge graphs sequentially based on each headline-attached text fragment data 511, and combines the sequentially generated first knowledge graphs to generate the text knowledge graph 600.

[0106] Specifically, the information processing device 100 generates a prompt requesting the generation of a first knowledge graph based on each of the headline-attached sentence fragment data 511, and inputs the generated prompt into a large-scale language model, thereby sequentially generating the first knowledge graphs. The large-scale language model is included in, for example, the generation processing device 202. The large-scale language model may be included in, for example, the information processing device 100. Specifically, the information processing device 100 generates a sentence knowledge graph 600 by combining the sequentially generated first knowledge graphs.

[0107] (6-2) The information processing device 100 stores the generated sentence knowledge graph 600 in a graph database 610. The graph database 610 is included in, for example, the information storage device 201. The graph database 610 may be included in, for example, the information processing device 100. In the graph database 610, the sentence knowledge graph 600 is represented by data 611. For example, the connection relationship between nodes is represented in the format (·)-[·]->(·). (·) represents a node. (·) is, for example, (:Task {name: "AAA"}). [·] represents a connection relationship. [·] is, for example, [:OTHER_CONNECT] or [:PART_OF].

[0108] (6-3) The information processing device 100 references the context information represented by the generated sentence knowledge graph 600 and generates an overall knowledge graph for the entire target data 500 based on one or more pieces of captioned table fragment data 512. Now, moving on to the explanation of Figs. 7 to 10, an example in which the information processing device 100 generates an overall knowledge graph will be described.

[0109] In FIG. 7, (7-1) the information processing device 100 generates a prompt requesting the generation of a second knowledge graph based on, for example, the sentence knowledge graph 600 and the first captioned table fragment data 512. The information processing device 100 generates the second knowledge graph by inputting the generated prompt into a large-scale language model. Now, moving on to the explanation of FIG. 8, an example of the prompt will be described. As shown in FIG. 8, the prompt requests reference to the sentence knowledge graph 600, requests the generation of a second knowledge graph, and includes data 800 specifying the format of the second knowledge graph.

[0110] Returning to the description of FIG. 7, (7-2) the information processing device 100 updates the sentence knowledge graph 600 by merging the generated second knowledge graph with the sentence knowledge graph 600. The information processing device 100 updates the sentence knowledge graph 600, for example, by updating the stored content of the graph database 610 from data 611 to data 700. In the example of FIG. 7, the bold text in the data 700 indicates the updated parts. Next, we move on to the description of FIG. 9.

[0111] 9, (9-1) the information processing device 100 generates a prompt requesting the generation of a second knowledge graph based on, for example, the latest sentence knowledge graph 600 and the second captioned table fragment data 512. The information processing device 100 generates the second knowledge graph by inputting the generated prompt into a large-scale language model.

[0112] (9-2) The information processing device 100 further updates the latest sentence knowledge graph 600 by merging the generated second knowledge graph with the latest sentence knowledge graph 600. The information processing device 100 updates the latest sentence knowledge graph 600, for example, by updating the stored contents of the graph database 610 from data 700 to data 900. In the example of FIG. 9, the bold text in the data 900 indicates the updated parts. Now, we move on to the explanation of FIG. 10 to explain an example of merging.

[0113] 10, it is assumed that there is a latest text knowledge graph 1000. The text knowledge graph 1000 is expressed, for example, by data 1001. Here, a case will be described in which the information processing device 100 generates second knowledge graphs 1011, 1012, etc. based on the captioned table fragment data 512 and updates the text knowledge graph 1000.

[0114] The sentence knowledge graph 1000 and the second knowledge graphs 1011 and 1012 each include a node representing a common entity "BBB" that has a common label type "Task." Therefore, when the information processing device 100 combines the sentence knowledge graph 1000 and the second knowledge graphs 1011 and 1012, it is preferable that the information processing device 100 integrates multiple nodes representing the common entity "BBB" into one node.

[0115] Therefore, the information processing device 100 updates the sentence knowledge graph 1000 to a sentence knowledge graph 1020. The sentence knowledge graph 1020 is expressed by, for example, data 1021. In this way, the information processing device 100 can connect node groups so that any node group is not isolated from other node groups. Therefore, the information processing device 100 can make the sentence knowledge graph 1020 more easily represent the relationships between entities appropriately.

[0116] Returning to the explanation of FIG. 9, (9-3) the information processing device 100 similarly generates second knowledge graphs for the third and subsequent headlined table fragment data 512, and performs update processing to further update the latest sentence knowledge graph 600. After performing update processing for the last headlined table fragment data 512, the information processing device 100 generates an overall knowledge graph by adopting the latest sentence knowledge graph 600 as the overall knowledge graph. This allows the information processing device 100 to acquire a useful overall knowledge graph.

[0117] The information processing device 100 can, for example, generate nodes and edges that ultimately form the entire knowledge graph in accordance with each of the headline-attached sentence fragment data 511. Similarly, the information processing device 100 can, for example, generate nodes and edges that ultimately form the entire knowledge graph in accordance with each of the headline-attached table fragment data 512. Therefore, the information processing device 100 can, for example, increase the number of nodes and edges that form the entire knowledge graph.

[0118] The information processing device 100 can generate an overall knowledge graph, for example, via headings, to appropriately represent the relationships between different entities that appear in different headline-attached sentence fragment data 511. Similarly, the information processing device 100 can generate an overall knowledge graph, for example, to appropriately represent the relationships between different entities that appear in different headline-attached table fragment data 512.

[0119] Furthermore, the information processing device 100 can generate an overall knowledge graph based on one or more pieces of table fragment data with headings 512 while taking into consideration, for example, context information represented by the sentence knowledge graph 600. This allows the information processing device 100 to appropriately detect relationships between entities that are difficult to detect using only the table fragment data with headings 512, and reflect these relationships in the overall knowledge graph. This allows the information processing device 100 to generate an overall knowledge graph that appropriately represents, for example, relationships between entities corresponding to table items.

[0120] (Specific example of operation of information processing device 100) Next, a specific example of the operation of the information processing device 100 will be described with reference to FIGS.

[0121] Fig. 11 is an explanatory diagram showing a specific example of the operation of the information processing device 100. In Fig. 11, (11-1) the information processing device 100 acquires document data 1100 representing a design document, development regulations, or the like, and a processing request requesting generation of a knowledge graph based on the document data 1100. The document data 1100 may be data corresponding to spreadsheet software. Now, moving on to the explanation of Figs. 12 and 13, an example of the document data 1100 will be described.

[0122] 12 and 13 are explanatory diagrams showing an example of document data 1100. As shown in FIGS. 12 and 13, the document data 1100 includes, for example, data 1200. As shown in FIG. 12, the data 1200 includes, for example, one or more headings and one or more sentences. Examples of the headings are "1.1 Operational Items" and "(1) Steady-State Operation." Examples of the sentences are "Steady-state operation is defined as the tasks and functions for safely and stably performing steady-state system operation daily, weekly, and monthly." Next, we move on to the explanation of FIG. 13.

[0123] As shown in FIG. 13, the data 1200 includes, for example, a table 1300. The table 1300 has fields for item number, item, content, and operation of the DR site when the main site is in operation. In the following explanation, the first row indicating the fields may be referred to as a "header." Furthermore, the second and subsequent rows may be referred to as "records." The item number field contains an item number that identifies the record. The item field contains the name of the item. The content field contains the specific content of the item. The field for operation of the DR site when the main site is in operation contains the content of the operation.

[0124] Returning to the description of FIG. 11, (11-2) in response to receiving a processing request, the information processing device 100 converts the document data 1100 into a Markdown document 1110 in accordance with Markdown notation. In Markdown notation, for example, a tag such as "###" is assigned to a heading. For example, a tag such as "-" is assigned to an entity such as a sentence. Now, moving on to the description of FIGS. 14 and 15, an example of the Markdown document 1110 will be described.

[0125] 14 and 15 are explanatory diagrams showing an example of a Markdown document 1110. As shown in Fig. 14 and 15, the Markdown document 1110 includes, for example, data 1400. As shown in Fig. 14, the data 1400 includes, for example, one or more headings tagged with tags such as "###" and one or more sentences tagged with tags such as "-". Next, we will move on to an explanation of Fig. 15.

[0126] 15, data 1400 includes, for example, data 1500 representing table 1300. Data 1500 represents the names of each field in the header, separated by "|". Data 1500 represents the values ​​of each field in each record, separated by "|".

[0127] Returning to the explanation of Fig. 11, (11-3) the information processing device 100 separates text data 1121 representing the text portion and table data 1122 representing the table portion from the Markdown document 1110. The text data 1121 includes, for example, the portion of the data 1400 shown in Fig. 14. The table data 1122 includes, for example, the data 1500 shown in Fig. 15 from the data 1400.

[0128] (11-4) The information processing device 100 fragments the text data 1121 to generate, from the text data 1121, a plurality of text fragment data 1131 and header data 1133 for the plurality of text fragment data 1131. The header data 1133 is list data that compiles one or more headers. The text fragment data 1131 is fragment data that represents any piece of text that belongs to any area in the Markdown document 1110 and is represented by any combination of one or more headers.

[0129] (11-5) The information processing device 100 fragments the table data 1122 to generate, from the table data 1122, a plurality of table fragment data 1132 and header data 1134 for the plurality of table fragment data 1132. The header data 1134 is list data that compiles one or more headers. The table fragment data 1132 is fragment data representing a header and one record for any table belonging to any area in the Markdown document 1110, represented by any combination of one or more headers. The table fragment data 1132 may represent two or more records for any table. If the plurality of table fragment data 1132 relate to the same table, each has a duplicated header.

[0130] (11-6) The information processing device 100 generates a plurality of heading-added sentence fragment data 1141 by adding one or more headings to each of the sentence fragment data 1131 based on the heading data 1133. For example, the information processing device 100 identifies a combination of one or more hierarchically structured headings in the Markdown document 1110 for the sentence fragment data 1131 that indicate the area to which the content of the sentence fragment data 1131 belongs, and adds the combination to the sentence fragment data 1131. Now, moving on to the description of FIG. 16 , an example of heading-added sentence fragment data 1141 will be described.

[0131] FIG. 16 is an explanatory diagram showing an example of heading-attached sentence fragment data 1141. As shown in FIG. 16, the heading-attached sentence fragment data 1141 includes, for example, data 1601 and 1602. Data 1601 specifically represents the sentence "This document (omitted) is not subject to operation." to which the heading "###1.1 Operation Items" is attached. Data 1602 specifically represents the sentence "Regarding normal operation (omitted) its contents shall be described." to which the heading "###1.1 Operation Items" and the heading "####(1) Normal Operation" are attached. As shown in FIG. 16, data 1602 includes the heading "####(1) Normal Operation" immediately above the sentence "Regarding normal operation (omitted) its contents shall be described." as well as the heading "####1.1 Operation Items," which is hierarchically higher.

[0132] Returning to the explanation of Fig. 11, (11-7) the information processing device 100 generates a plurality of table fragment data 1142 with headings by assigning one or more headings to each table fragment data 1132 based on the heading data 1134. For example, the information processing device 100 identifies a combination of one or more hierarchically structured headings that indicate an area to which the content of the table fragment data 1132 belongs in the Markdown document 1110 for the table fragment data 1132, and assigns the combination to the table fragment data 1132. Now, moving on to the explanation of Fig. 17, an example of table fragment data 1142 with headings will be described.

[0133] FIG. 17 is an explanatory diagram showing an example of heading-attached table fragment data 1142. As shown in FIG. 17, heading-attached table fragment data 1142 includes, for example, data 1701 to 1704. Data 1701 specifically represents a table record "|Item number|Item|(omitted)|Implemented in the same way as main site|" to which headings "####1.1 Operation items" and "####(1) Normal operation" are assigned. As shown in FIG. 17, data 1701 includes the heading "####(1) Normal operation" immediately above the table record "|Item number|Item|(omitted)|Implemented in the same way as main site|", as well as the heading "####1.1 Operation items" which is hierarchically higher.

[0134] Specifically, data 1702 represents the table record "|Item number|Item|(omitted)|Implemented in the same way as the main site|" with the heading "####1.1 Operation items" and the heading "####(1) Normal operation." As shown in Fig. 17, data 1702 includes the heading "####(1) Normal operation" immediately above the table record "|Item number|Item|(omitted)|Implemented in the same way as the main site|," as well as the heading "####1.1 Operation items," which is hierarchically higher.

[0135] Specifically, data 1703 represents the table record "|Item number|Item|(omitted)|Implemented in the same way as the main site|" with the heading "####1.1 Operation items" and the heading "####(1) Normal operation." As shown in Fig. 17, data 1703 includes the heading "####(1) Normal operation" immediately above the table record "|Item number|Item|(omitted)|Implemented in the same way as the main site|," as well as the heading "####1.1 Operation items," which is hierarchically higher.

[0136] Specifically, data 1704 represents the table record "|Item number|Item|(omitted)|Implemented in the same way as the main site|" with the heading "####1.1 Operation items" and the heading "####(1) Normal operation." As shown in Fig. 17, data 1704 includes the heading "####(1) Normal operation" immediately above the table record "|Item number|Item|(omitted)|Implemented in the same way as the main site|," as well as the heading "####1.1 Operation items," which is hierarchically higher.

[0137] If the document data 1100 includes headings such as "1.2 Operation Items" and "(2) Operation During Failure," the heading-attached table fragment data 1142 may be, for example, data 1705. Specifically, data 1705 represents the table record "|Item Number|Item|(Omitted)|Implemented in the same way as the main site|" to which the headings "####1.2 Operation Items" and "####(2) Operation During Failure" are assigned. As shown in FIG. 17, data 1705 includes the heading "####(2) Operation During Failure" immediately above the table record "|Item Number|Item|(Omitted)|Implemented in the same way as the main site|," as well as the hierarchically higher heading "####1.2 Operation Items."

[0138] Returning to the explanation of Fig. 11, (11-8) the information processing device 100 generates sentence graph data 1151 representing a sentence knowledge graph based on a plurality of headline-attached sentence fragment data 1141. The information processing device 100 generates the sentence graph data 1151 based on a plurality of headline-attached sentence fragment data 1141 by utilizing a large-scale language model.

[0139] (11-9) The information processing device 100 generates overall graph data 1152 representing an overall knowledge graph based on the sentence graph data 1151 and a plurality of captioned table fragment data 1142. The information processing device 100 uses a large-scale language model to generate the overall graph data 1152 based on the sentence graph data 1151 and a plurality of captioned table fragment data 1142. This allows the information processing device 100 to acquire the overall graph data 1152 representing a useful overall knowledge graph.

[0140] The information processing device 100 can, for example, generate nodes and edges that ultimately form the entire knowledge graph in accordance with each piece of headline-attached sentence fragment data 1141. Similarly, the information processing device 100 can, for example, generate nodes and edges that ultimately form the entire knowledge graph in accordance with each piece of headline-attached table fragment data 1142. Therefore, the information processing device 100 can, for example, increase the number of nodes and edges that form the entire knowledge graph.

[0141] The information processing device 100 can generate an overall knowledge graph, for example, via headings, to appropriately represent the relationships between different entities that appear in different headline-attached sentence fragment data 1141. Similarly, the information processing device 100 can generate an overall knowledge graph, for example, via headings, to appropriately represent the relationships between different entities that appear in different headline-attached table fragment data 1142. Specifically, because headings serve as nodes, the information processing device 100 can easily generate a useful overall knowledge graph simply by adding headings, as in the headline-attached sentence fragment data 1141 and the headline-attached table fragment data 1142.

[0142] Furthermore, the information processing device 100 can generate an overall knowledge graph based on a plurality of captioned table fragment data 1142 while taking into consideration, for example, context information represented by the sentence knowledge graph. Therefore, the information processing device 100 can appropriately detect relationships between entities that are difficult to detect using only the captioned table fragment data 1142, and can reflect these relationships in the overall knowledge graph. The information processing device 100 can appropriately detect relationships between entities by utilizing, for example, context information from the sentence knowledge graph, which is considered to have a relatively large amount of information. Therefore, the information processing device 100 can generate an overall knowledge graph that appropriately represents relationships between entities corresponding to, for example, table items.

[0143] Here, a case has been described in which the information processing device 100 generates a sentence knowledge graph based on a plurality of headlined sentence fragment data 1141, and then generates an entire knowledge graph based on a plurality of headlined table fragment data 1142, but this is not limiting. For example, the information processing device 100 may generate a table knowledge graph based on a plurality of headlined table fragment data 1142, and then generate an entire knowledge graph based on a plurality of headlined sentence fragment data 1141. A table knowledge graph represents, for example, relationships between entities that appear in a table.

[0144] Specifically, the information processing device 100 generates a table knowledge graph based on a plurality of headline-attached table fragment data 1142. Specifically, the information processing device 100 generates an entire knowledge graph based on the generated table knowledge graph and a plurality of headline-attached sentence fragment data 1141. This allows the information processing device 100 to generate the entire knowledge graph based on the plurality of headline-attached sentence fragment data 1141 while taking into consideration the context information represented by the table knowledge graph. Therefore, the information processing device 100 can generate the entire knowledge graph so as to appropriately represent the relationships between entities.

[0145] Here, a case has been described in which the information processing device 100 assigns one or more headings to each piece of sentence fragment data 1131, but this is not limiting. For example, the information processing device 100 may assign one or more headings to each piece of sentence fragment data 1131 together with the sentences that exist directly under each heading.

[0146] Here, a case has been described in which the information processing device 100 assigns one or more headings to each table fragment data 1132, but this is not limiting. For example, there may be a case in which the information processing device 100 assigns one or more headings to each table fragment data 1132 together with text that exists directly under each heading.

[0147] (Example of effect of information processing device 100) Next, an example of the effect achieved by the information processing device 100 will be described with reference to FIGS.

[0148] 18 to 20 are explanatory diagrams showing an example of the effect. In Fig. 18, a knowledge graph 1800 generated by a conventional method for target data is compared with an overall knowledge graph 1810 generated by the information processing device 100. The conventional method is a method of dividing target data into multiple fragments, for example, and generating an overall knowledge graph by collecting partial knowledge graphs based on each fragment.

[0149] As shown in knowledge graph 1800, each of node groups 1801 to 1805 exists independently and is not connected to each of the other node groups 1801 to 1805. Node group 1801 relates to, for example, the item "system monitoring operation." Node group 1802 relates to, for example, the item "batch operation." Node group 1803 relates to, for example, the item "operation schedule." Node group 1804 relates to, for example, the item "failure response." Node group 1805 relates to, for example, the item "media management operation method." Therefore, with conventional methods, knowledge graph 1800 cannot represent the relationships between entities belonging to different items, and is therefore considered to be unuseful.

[0150] In contrast, as shown in the overall knowledge graph 1810, each node is directly or indirectly connected via node 1811 using headings. Node 1811 corresponds to the heading "Normal Operation," for example. Therefore, the overall knowledge graph 1810 can appropriately represent the relationships between entities belonging to different items, thereby improving the probability of usefulness. In this way, the information processing device 100 can make the useful overall knowledge graph 1810 available. Next, we move on to the description of FIG. 19.

[0151] In FIG. 19, graphs 1900 and 1910 show an example of an overall knowledge graph when the information processing device 100 generates an overall knowledge graph after generating a table knowledge graph. In this case, the number of nodes forming the overall knowledge graph is 178. The number of edges forming the overall knowledge graph is 207. Therefore, compared to conventional methods, the information processing device 100 can generate a useful overall knowledge graph formed of a relatively large number of nodes and edges. Next, we move on to the explanation of FIG. 20.

[0152] 20, graphs 2000 and 2010 show an example of an overall knowledge graph when the information processing device 100 generates an overall knowledge graph after generating a sentence knowledge graph. In this case, the number of nodes forming the overall knowledge graph is 189. The number of edges forming the overall knowledge graph is 229. Therefore, the information processing device 100 can generate a useful overall knowledge graph that is formed with even more nodes and even more edges than when generating a table knowledge graph.

[0153] (Overall processing procedure) Next, an example of an overall processing procedure executed by the information processing device 100 will be described with reference to Fig. 21. The overall processing is realized by, for example, the CPU 301, storage areas such as the memory 302 and the recording medium 305, and the network I / F 303 shown in Fig. 3.

[0154] 21 is a flowchart showing an example of an overall processing procedure, in which the information processing device 100 generates a Markdown document based on a target document such as a design document or a development agreement (step S2101).

[0155] Next, the information processing device 100 divides the Markdown document into a text portion and a table portion, and generates text data corresponding to the divided text portion and table data corresponding to the divided table portion (step S2102).

[0156] Then, the information processing device 100 fragments the text data to generate one or more text fragment data (step S2103). Also, the information processing device 100 fragments the table data to generate one or more table fragment data (step S2104). Also, the information processing device 100 identifies one or more headings (step S2105).

[0157] Next, the information processing device 100 generates one or more headline-attached sentence fragment data by assigning a headline corresponding to each of the one or more identified sentence fragment data to the sentence fragment data (step S2106).

[0158] In addition, the information processing device 100 generates one or more table fragment data with headings by assigning headings corresponding to each of the one or more table fragment data from among the one or more identified headings to the table fragment data (step S2107).

[0159] Next, the information processing device 100 generates sentence graph data representing a knowledge graph related to the sentence based on the one or more generated headline sentence fragment data (step S2108).Then, the information processing device 100 generates overall graph data representing a knowledge graph related to the target document based on the generated sentence graph data and the one or more generated headline table fragment data (step S2109).

[0160] Next, the information processing device 100 outputs the entire graph data (step S2110). Then, the information processing device 100 ends the entire process. As a result, the information processing device 100 can make the entire graph data representing a useful knowledge graph related to the target document available externally.

[0161] Here, the information processing device 100 may change the order of the processes of some of the steps in the flowchart of Fig. 21. Furthermore, the information processing device 100 may omit the processes of some of the steps in the flowchart of Fig. 21.

[0162] (Application example of information processing device 100) The information processing device 100 can be applied to, for example, applications that support workers who create test specifications. Specifically, the information processing device 100 generates an overall knowledge graph based on design documents, development rules, or the like. Specifically, the worker uses the overall knowledge graph to search design information, identify comprehensive test items, and create test specifications. This allows the information processing device 100 to reduce the workload of the worker. Furthermore, the information processing device 100 can make it easier to create useful test specifications.

[0163] As described above, the information processing device 100 can acquire first information about sentences included in target data. The information processing device 100 can acquire second information about graphic elements included in target data. The information processing device 100 can acquire third information about headings included in target data. The information processing device 100 can select either the acquired first information or the acquired second information. The information processing device 100 can identify a heading that corresponds to one of the selected pieces of information among headings appearing in the target data. The information processing device 100 can generate a first knowledge graph about the target data based on one of the selected pieces of information and the identified heading that corresponds to one of the selected pieces of information. The information processing device 100 can select the other of the acquired first information and the acquired second information. The information processing device 100 can identify a heading that corresponds to the other of the selected pieces of information among headings by referring to the acquired third information. According to the information processing device 100, it is possible to generate a second knowledge graph related to the target data based on the other selected one, the identified heading corresponding to the information of the other selected one, and the generated first knowledge graph. This allows the information processing device 100 to generate a useful second knowledge graph.

[0164] According to the information processing device 100, it is possible to select the first information as one of the pieces of information. According to the information processing device 100, it is possible to select the second information as the other piece of information. This makes it easier for the information processing device 100 to generate a more useful second knowledge graph.

[0165] The information processing device 100 can process a plurality of first information, a plurality of second information, a plurality of third information, and a plurality of headings. The information processing device 100 can assign, to each piece of acquired first information, a first heading among the headings that corresponds to the first information, based on the acquired third information. The information processing device 100 can assign, to each piece of acquired second information, a second heading among the headings that corresponds to the second information, based on the acquired third information. The information processing device 100 can generate a first knowledge graph based on either the first information to which the first headings have been assigned or the second information to which the second headings have been assigned. The information processing device 100 can generate a second knowledge graph based on the other of the first information to which the first headings have been assigned or the second information to which the second headings have been assigned. This allows the information processing device 100 to generate a second knowledge graph by collectively treating pairs of first information and first headings or pairs of second information and second headings. This allows the information processing device 100 to easily generate a useful second knowledge graph that takes headlines into consideration.

[0166] The information processing device 100 can utilize a language model that enables generation of a knowledge graph in response to an input prompt. The information processing device 100 can generate a first knowledge graph by inputting a prompt to the language model requesting generation of a first knowledge graph. The information processing device 100 can generate a second knowledge graph by inputting a prompt to the language model requesting generation of a second knowledge graph based on the first knowledge graph. This allows the information processing device 100 to use the language model to facilitate generation of the second knowledge graph.

[0167] According to the information processing device 100, it is possible to identify, as a first heading corresponding to the first information, each of the hierarchically structured headings that form a combination of headings that indicate a data area to which the first information belongs in the target data, among the headings having a hierarchical structure. According to the information processing device 100, it is possible to identify, as a second heading corresponding to the second information, each of the hierarchically structured headings that form a combination of headings that indicate a data area to which the second information belongs in the target data, among the headings having a hierarchical structure. This makes it possible for the information processing device 100 to easily generate a useful second knowledge graph that appropriately represents the relationships between entities, using appropriate headings as guidelines.

[0168] According to the information processing device 100, it is possible to generate first information relating to different sentences among a plurality of sentences, each of which has a predetermined length, that are divided from the target data. According to the information processing device 100, it is possible to generate second information relating to different records among a plurality of records that form a table that is divided from the target data. This allows the information processing device 100 to prepare the first information and the second information without relying on another computer.

[0169] The information processing method described in this embodiment can be realized by executing a prepared program on a computer such as a PC or a workstation. The information processing program described in this embodiment is recorded on a computer-readable recording medium and executed by being read from the recording medium by the computer. The recording medium may be a hard disk, a flexible disk, a CD (Compact Disc)-ROM, an MO (Magneto Optical disc), a DVD (Digital Versatile Disc), or the like. The information processing program described in this embodiment may also be distributed via a network such as the Internet.

[0170] The following additional notes are provided regarding the above-described embodiment.

[0171] (Supplementary Note 1) Acquire first information about sentences included in the target data, second information about graphic elements included in the target data, and third information about headings included in the target data; generating a first knowledge graph for the target data based on either the acquired first information or the acquired second information and a heading corresponding to the one piece of information among headings included in the target data, with reference to the acquired third information; referencing the acquired third information, and generating a second knowledge graph related to the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other information, and the generated first knowledge graph; An information processing program that causes a computer to execute a process.

[0172] (Supplementary Note 2) The one of the two is the first information, The information processing program according to Supplementary Note 1, wherein the other information is the second information.

[0173] (Supplementary Note 3) The first information, the second information, the third information, and the heading are each plural, Based on the acquired third information, assigning a first heading from among the headings corresponding to the first information to each of the acquired first information, and assigning a second heading from among the headings corresponding to the second information to each of the acquired second information; causing the computer to execute a process; The process of generating the first knowledge graph includes: generating the first knowledge graph based on either the first information to which the first headings are assigned and the second information to which the second headings are assigned; The process of generating the second knowledge graph includes: The information processing program according to claim 2, wherein the second knowledge graph is generated based on either the first information to which the first heading is assigned, or the second information to which the second heading is assigned.

[0174] (Supplementary Note 4) The process of generating the first knowledge graph includes: generating the first knowledge graph by inputting a prompt to a language model that can generate a knowledge graph in response to an input prompt, the prompt including either the first information to which the first heading is assigned or the second information to which the second heading is assigned, and requesting generation of the first knowledge graph based on the either information; The process of generating the second knowledge graph includes: and generating the second knowledge graph by inputting a prompt to the language model, the prompt requesting that the language model generate the second knowledge graph based on the first knowledge graph and the other of the first information to which the first heading is assigned and the second information to which the second heading is assigned.

[0175] (Appendix 5) The headings have a hierarchical structure, The first heading corresponding to the first information is a heading that forms a combination of headings among the headings, indicating a data area to which the first information in the target data belongs, The information processing program described in Appendix 4, characterized in that the second heading corresponding to the second information is a heading among the headings that forms a combination of headings indicating the data area to which the second information belongs in the target data.

[0176] (Appendix 6) The acquisition process is An information processing program described in any one of Appendices 1 to 5, characterized in that it generates the first information regarding different sentences from a plurality of sentences, each of which has a predetermined length, that are divided from the target data, and generates the second information regarding different records from a plurality of records that form a table that is divided from the target data.

[0177] (Appendix 7) Acquire first information about sentences included in the target data, second information about graphic elements included in the target data, and third information about headings included in the target data; generating a first knowledge graph for the target data based on either the acquired first information or the acquired second information and a heading corresponding to the one piece of information among headings included in the target data, with reference to the acquired third information; referencing the acquired third information, and generating a second knowledge graph related to the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other information, and the generated first knowledge graph; An information processing method characterized in that the processing is executed by a computer.

[0178] (Appendix 8) Acquire first information about sentences included in the target data, second information about graphic elements included in the target data, and third information about headings included in the target data; generating a first knowledge graph for the target data based on either the acquired first information or the acquired second information and a heading corresponding to the one piece of information among headings included in the target data, with reference to the acquired third information; referencing the acquired third information, and generating a second knowledge graph related to the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other information, and the generated first knowledge graph; An information processing device comprising a control unit. [Explanation of symbols]

[0179] 100 Information processing device 110,500 target data 111 1st information 112 Second information 113 Third information 121 First Knowledge Graph 122,1011,1012 Second Knowledge Graph 200 Information Processing Systems 201 Information storage device 202 Generation processing device 203 Client Device 210 Network 300 Bus 301 CPU 302 memory 303 Network I / F 304 Recording Media I / F 305 Recording Media 306 Display 307 Input Device 400 Storage section 401 Acquisition Department 402 Split part 403 1st generation part 404 Second generation part 405 Output Section 501,511,1131,1141 Text fragment data 502,512,1132,1142 Table fragment data 503, 611, 700, 800, 900, 1001, 1021, 1133, 1134, 1200, 1400, 1500, 1601, 1602, 1701-1705 data 600, 1000, 1020 sentence knowledge graph 610 Graph Database 1100 Document Data 1110 Markdown Document 1121 Text data 1122 Table Data 1151 Text graph data 1152 Overall graph data 1300 tables 1800 Knowledge Graph 1801~1805 node group 1810 Overall Knowledge Graph 1811 nodes 1900,1910,2000,2010 graph

Claims

1. acquiring first information about sentences included in the target data, second information about graphic elements included in the target data, and third information about headings included in the target data; generating a first knowledge graph for the target data based on either the acquired first information or the acquired second information and a heading corresponding to the one piece of information among headings included in the target data, with reference to the acquired third information; referencing the acquired third information, and generating a second knowledge graph related to the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other information, and the generated first knowledge graph; An information processing program that causes a computer to execute a process.

2. The one is the first information, 2. The information processing program according to claim 1, wherein the other information is the second information.

3. the first information, the second information, the third information, and the heading are each plural; Based on the acquired third information, a first heading corresponding to the first information is assigned to each of the acquired first information, and a second heading corresponding to the second information is assigned to each of the acquired second information. causing the computer to execute a process; The process of generating the first knowledge graph includes: generating the first knowledge graph based on either the first information to which the first headings are assigned and the second information to which the second headings are assigned; The process of generating the second knowledge graph includes:

3. The information processing program according to claim 2, wherein the second knowledge graph is generated based on either the first information to which the first headings are assigned or the second information to which the second headings are assigned.

4. The process of generating the first knowledge graph includes: generating the first knowledge graph by inputting a prompt to a language model that can generate a knowledge graph in response to an input prompt, the prompt including either the first information to which the first heading is assigned or the second information to which the second heading is assigned, and requesting generation of the first knowledge graph based on the either information; The process of generating the second knowledge graph includes:

4. The information processing program according to claim 3, wherein the second knowledge graph is generated by inputting a prompt to the language model requesting that the second knowledge graph be generated based on the first knowledge graph and the other of the first information to which the first heading is assigned and the second information to which the second heading is assigned.

5. The headings have a hierarchical structure, The first heading corresponding to the first information is a heading that forms a combination of headings among the headings and indicates a data area to which the first information in the target data belongs, The information processing program according to claim 4, characterized in that the second heading corresponding to the second information is each heading among the headings that forms a combination of headings that indicates the data area to which the second information belongs in the target data.

6. acquiring first information about sentences included in the target data, second information about graphic elements included in the target data, and third information about headings included in the target data; generating a first knowledge graph for the target data based on either the acquired first information or the acquired second information and a heading corresponding to the one piece of information among headings included in the target data, with reference to the acquired third information; referencing the acquired third information, and generating a second knowledge graph related to the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other information, and the generated first knowledge graph; An information processing method characterized in that the processing is executed by a computer.

7. acquiring first information about sentences included in the target data, second information about graphic elements included in the target data, and third information about headings included in the target data; generating a first knowledge graph for the target data based on either the acquired first information or the acquired second information and a heading corresponding to the one piece of information among headings included in the target data, with reference to the acquired third information; referencing the acquired third information, and generating a second knowledge graph related to the target data based on the other of the acquired first information and the acquired second information, a heading among the headings that corresponds to the other information, and the generated first knowledge graph; An information processing device comprising a control unit.

Images