Data linkage device, data linkage system, data linkage method, and data linkage program
Patent Information
- Application Number
- PCT/JP2025/018530
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2025-02-20
- Filing Date
- 2025-05-22
- Publication Date
- 2026-08-27
Smart Images
Figure JP2025018530_27082026_PF_FP_ABST
Abstract
Description
Data Linking Device, Data Linking System, Data Linking Method, and Data Linking Program
[0001] The present disclosure relates to a data linking device, a data linking system, a data linking method, and a data linking program.
[0002] As a technique for realizing data linking, vector search is used. For example, data linking in an RDB or ontology can be realized by searching for columns or nodes with the same or similar names. At this time, the cosine similarity is often used as an index of similarity. The magnitude of the numerical value of this cosine similarity serves as the basis for determining whether data linking is possible.
[0003] In Patent Document 1, an attribute value of a column name in a table format is combined into one character group. Then, a technique for performing feature extraction and similarity comparison on the character group is disclosed.
[0004] Japanese Patent Application Laid-Open No. 2020-112919
[0005] In the vector search as used in Patent Document 1, the properties of an entity that is the target of data linking are represented in a vector space. Therefore, from the human perspective, the properties of an entity are a black box. Thus, humans only know that the cosine similarity between entities capable of data linking is large. As described above, there is a problem in vector search that the basis of data linking cannot be shown in a form understandable to humans.
[0006] An object of the present disclosure is to provide a data linking device that can show the properties of an entity that is the basis of data linking in a form understandable to humans.
[0007] The data linkage device according to this disclosure includes: a property extraction unit that acquires source entity information, which is information associated with the source entity in data linkage, and a viewpoint representing the source entity; generates a first prompt for generating source property information, which is a property representing the source entity using the viewpoint, from the source entity information; and inputs the first prompt to artificial intelligence to extract the source property information from the source entity information; and a linkage determination unit that uses the source property information and the destination entity information, which is information associated with the destination entity in data linkage, to determine whether data linkage between the source entity and the destination entity is possible.
[0008] The data linkage device described in this disclosure can provide a data linkage device that can display the properties of the source entity that forms the basis of the data linkage in a human-understandable format.
[0009] A diagram showing an example of the overall configuration of the data linkage system according to Embodiment 1. A diagram showing an example of the configuration of the data linkage device according to Embodiment 1. A diagram showing an example of the functional configuration of the data linkage device according to Embodiment 1. A flowchart showing an example of the processing of the data linkage process according to Embodiment 1. A schematic diagram showing an example of the processing of the data linkage process according to Embodiment 1. A schematic diagram showing an example of the processing of acquiring linked source property information according to Embodiment 1. A schematic diagram showing an example of the processing of property extraction process according to Modification 1 of Embodiment 1. A diagram showing an example of the functional configuration of the data linkage device according to Modification 2 of Embodiment 1. A flowchart showing an example of the processing of the data linkage process according to Modification 2 of Embodiment 1. A schematic diagram showing an example of the processing of the data linkage process according to Modification 2 of Embodiment 1. A diagram showing an example of the configuration of the data linkage device according to Modification 4 of Embodiment 1. A schematic diagram explaining the effects of the data linkage system according to Embodiment 1. A diagram showing an example of the functional configuration of the data linkage device according to Embodiment 2. A flowchart showing an example of the processing of the data linkage process according to Embodiment 2. A schematic diagram showing an example of the processing of the data linkage process according to Embodiment 2. A diagram showing an example of the functional configuration of the data linkage device according to Embodiment 3. A flowchart showing an example of the processing of the data linkage process according to Embodiment 3. A schematic diagram showing an example of data linkage processing according to Embodiment 3.
[0010] The following description of this embodiment will be illustrated with reference to the figures. In each figure, identical or corresponding parts are denoted by the same reference numerals. In the description of the embodiment, descriptions of identical or corresponding parts will be omitted or simplified as appropriate. The arrows in the figures mainly indicate the flow of data or processing. Also, the size relationships of the components in the following figures may differ from those in reality. In addition, in the description of the embodiment, directions or positions such as up, down, left, right, front, back, front, and back may be indicated. These notations are for the convenience of explanation and do not limit the arrangement, direction, and orientation of the devices, instruments, or parts.
[0011] Embodiment 1. ***Description of Configuration*** Figure 1 is a diagram showing an example of the overall configuration of the data linkage system 500 according to this embodiment. Figure 2 is a diagram showing an example of the configuration of the data linkage device 100 according to this embodiment.
[0012] The data linkage system 500 comprises artificial intelligence 200 and a data linkage device 100 that determines whether data linkage between a source entity and a destination entity is possible. The data linkage device 100 and the artificial intelligence 200 are capable of inputting and outputting data, or sending and receiving data.
[0013] First, an example configuration of the data linkage device 100 will be described. The data linkage device 100 is a computer. The data linkage device 100 includes a processor 910, as well as other hardware such as memory 921, auxiliary storage device 922, input interface 930, output interface 940, and communication device 950. The processor 910 is connected to the other hardware via signal lines and controls this other hardware.
[0014] The data linkage device 100 includes, as functional elements, a property extraction unit 110, a linkage determination unit 120, and a storage unit 150. The storage unit 150 stores the source entity database 511, the destination entity database 512, the source property database 521, the destination property database 522, and the linkage result database 530. In the diagram, databases are abbreviated as DB.
[0015] The functions of the property extraction unit 110 and the cooperation determination unit 120 are implemented by software. The storage unit 150 is provided in the memory 921. The storage unit 150 may also be provided in the auxiliary storage device 922, or it may be distributed between the memory 921 and the auxiliary storage device 922.
[0016] The processor 910 is a device that executes a data linkage program. The data linkage program is a program that realizes the functions of the property extraction unit 110 and the linkage determination unit 120. The processor 910 is an IC that performs arithmetic processing. Specific examples of the processor 910 are a CPU, DSP, and GPU. IC is an abbreviation for Integrated Circuit. CPU is an abbreviation for Central Processing Unit. DSP is an abbreviation for Digital Signal Processor. GPU is an abbreviation for Graphics Processing Unit.
[0017] Memory 921 is a storage device that temporarily stores data. Specific examples of memory 921 include SRAM or DRAM. SRAM is an abbreviation for Static Random Access Memory. DRAM is an abbreviation for Dynamic Random Access Memory. Auxiliary storage device 922 is a storage device that stores data. Specific examples of auxiliary storage device 922 include HDD. Alternatively, auxiliary storage device 922 may be a portable storage medium such as an SD® memory card, CF, NAND flash, flexible disk, optical disk, compact disk, Blu-ray® disc, or DVD. HDD is an abbreviation for Hard Disk Drive. SD® is an abbreviation for Secure Digital. CF is an abbreviation for CompactFlash (registered trademark). DVD is an abbreviation for Digital Versatile Disk.
[0018] The input interface 930 is a port to which input devices such as a mouse, keyboard, or touch panel are connected. Specifically, the input interface 930 is a USB terminal. The input interface 930 may also be a port connected to a LAN. USB is an abbreviation for Universal Serial Bus. LAN is an abbreviation for Local Area Network.
[0019] The output interface 940 is a port to which the cable of an output device, such as a display, is connected. Specifically, the output interface 940 is a USB terminal or an HDMI® terminal. Specifically, the display is an LCD. The output interface 940 is also called the display interface. HDMI® is an abbreviation for High Definition Multimedia Interface. LCD is an abbreviation for Liquid Crystal Display.
[0020] The communication device 950 has a receiver and a transmitter. The communication device 950 is connected to a communication network such as a LAN, the Internet, a telephone line, or Wi-Fi (registered trademark). Specifically, the communication device 950 is a communication chip or NIC. NIC is an abbreviation for Network Interface Card.
[0021] The data linkage program is executed in the data linkage device 100. The data linkage program is loaded into the processor 910 and executed by the processor 910. Memory 921 stores not only the data linkage program but also the OS. OS is an abbreviation for Operating System. The processor 910 executes the data linkage program while executing the OS. The data linkage program and the OS may also be stored in the auxiliary storage device 922. The data linkage program and OS stored in the auxiliary storage device 922 are loaded into memory 921 and executed by the processor 910. Note that part or all of the data linkage program may be incorporated into the OS.
[0022] The data linkage device 100 may include multiple processors that replace the processor 910. These multiple processors share the task of executing the data linkage program. Each processor is a device that executes the data linkage program, just like the processor 910.
[0023] The data, information, signal values, and variable values used, processed, or output by the data linkage program are stored in memory 921, auxiliary storage device 922, or registers or cache memory within the processor 910.
[0024] The word "part" in "property extraction unit 110" and "linking determination unit 120" may be replaced with "circuit," "process," "procedure," "process," or "circuitry." The data linking program causes a computer to execute the property extraction process and the linking determination process. The word "process" in "property extraction process" and "linking determination process" may be replaced with "program," "program product," "computer-readable storage medium storing the program," or "computer-readable recording medium recording the program." The data linking method is performed by the data linking device 100 executing the data linking program. The data linking program may be provided stored on a computer-readable recording medium. The data linking program may also be provided as a program product.
[0025] The artificial intelligence 200 includes an inference unit that performs inference using a trained model. The inference unit takes, for example, a first prompt 61 or a second prompt 62 from the property extraction unit 110 as input and outputs source property information 51 or destination property information 52 corresponding to the first prompt 61 based on the trained model. The artificial intelligence 200 may also be a generative AI. AI is an abbreviation for Artificial Intelligence. The trained model is also called a large language model, or LLM. LLM is an abbreviation for Large Language Model. The property extraction unit 110 inputs the first prompt 61 or the second prompt 62 to the external artificial intelligence 200 and instructs it to generate source property information 51 or destination property information 52. The property extraction unit 110 then outputs the generated source property information 51 or destination property information 52. In this way, various external artificial intelligence services can be used to output the source property information 51 or the destination property information 52. In Figure 1, the artificial intelligence 200 is an external component of the data linkage device 100, but it may also be an internal component.
[0026] The artificial intelligence 200 may be constructed using algorithms such as Transformer, BERT, and GPT. Alternatively, the artificial intelligence 200 may be constructed by combining multiple algorithms, including these algorithms. BERT is an abbreviation for Bidirectional EncoderRepresentations from Transformers. GPT is an abbreviation for Generative Pre-Training.
[0027] ***Explanation of Operation*** Next, the operation of the data linkage system 500 according to this embodiment will be explained. The operation procedure of the data linkage system 500 corresponds to the data linkage method. The program that realizes the data linkage processing, which is the operation of the data linkage system 500, corresponds to the data linkage program.
[0028] Figure 3 shows an example of the functional configuration of the data linkage device 100 according to this embodiment. The data linkage processing by the data linkage device 100 includes property extraction processing by the property extraction unit 110 and linkage determination processing. In the property extraction processing, the property extraction unit 110 acquires source entity information 21, which is information associated with the source entity, and a viewpoint 30 that represents the source entity. The property extraction unit 110 generates a first prompt 61 for generating source property information 51, which is a property representing the source entity, from the source entity information 21 using the viewpoint 30. Then, the property extraction unit 110 inputs the first prompt 61 to the artificial intelligence 200 to extract the source property information 51 from the source entity information 21. In this embodiment, the property extraction unit 110 further acquires destination entity information 22, which is information associated with the destination entity. The property extraction unit 110 generates a second prompt 62 for generating linked property information 52, which is a property representing the linked entity using the viewpoint 30, from the linked entity information 22. The property extraction unit 110 then inputs the second prompt 62 to the artificial intelligence 200 to extract the linked property information 52.
[0029] In the linking determination process, the linking determination unit 120 uses the source property information 51 and the destination entity information 22, which is information associated with the destination entity, to determine whether data linking between the source entity and the destination entity is possible. In this embodiment, the linking determination unit 120 uses the source property information 51 and the destination property information 52 to determine whether data linking between the source entity and the destination entity is possible.
[0030] Figure 4 is a flowchart showing an example of the data linkage process according to this embodiment. The data linkage process according to this embodiment will be described in detail.
[0031] <Property extraction process: Steps S101 to S111> The property extraction unit 110 obtains the source entity information 21, which is information associated with the source entity, the destination entity information 22, which is information associated with the destination entity, and the viewpoint 30. Specifically, it is as follows.
[0032] In step S101, the property extraction unit 110 obtains the viewpoint 30 entered by the user. The viewpoint 30 is a viewpoint that represents an entity and becomes a property item. In step S102, the property extraction unit 110 obtains an entity name set I, which is a set of source entity names i. Let n be the number of entity names i in the entity name set I. Let n be a natural number. An entity is the object of data linkage. A source entity is the object of linkage in data linkage. In step S103, the property extraction unit 110 obtains an entity name set J, which is a set of destination entity names j. Let m be the number of entity names j in the entity name set J. Let m be a natural number. A destination entity is the object of linkage in data linkage.
[0033] The process from step S104 to step S107 is to extract the source property information 51 from the viewpoint 30 and the source entity information 21. The process from step S108 to step S111 is to extract the destination property information 52 from the viewpoint 30 and the destination entity information 22.
[0034] In step S104, the property extraction unit 110 obtains source entity information 21 corresponding to the source entity name i from the source entity database 511. The source entity database 511 is a database that stores information (text, graphs, tables) associated with the entity name i.
[0035] Figure 5 is a schematic diagram showing an example of the data linkage process according to this embodiment. In Figure 5, "apple" is shown as an example element of the source entity name set I, and "peach" and "tomato" are shown as examples of elements of the destination entity name set J. In addition, a sentence describing apple is shown as the source entity information 21, which is information associated with the source entity "apple". A sentence describing peach is shown as the destination entity information 22, which is information associated with the destination entity "peach". A sentence describing tomato is shown as the destination entity information 22, which is information associated with the destination entity "tomato". Furthermore, in Figure 5, "classification" is shown as a perspective 30 for representing the entities to be linked, which is entered by the user.
[0036] In step S105, the property extraction unit 110 generates a first prompt 61 to obtain the source property information 51, which is a property representing the source entity using the viewpoint 30, from the source entity information 21. The property extraction unit 110 then inputs the first prompt 61 to the artificial intelligence 200 to extract the source property information 51 from the source entity information 21.
[0037] Figure 6 is a schematic diagram showing an example of the process for acquiring linked property information 51 according to this embodiment. Figure 6 shows the process of extracting the properties of an apple, "Classification: Rosaceae, Malus genus, Shape: Round, Taste: Sweet," from the linked entity information 21, "Text describing an apple." The property extraction unit 110 obtains "#Specified property: Classification, Shape, Taste" from the user as the perspective 30. The perspective 30 is also called the specified property. The property extraction unit 110 accepts "#Entity: Apple" from the user as the name of the linked entity. The property extraction unit 110 also accepts "#Text: Text describing an apple" from the user as the linked entity information 21. The property extraction unit 110 then generates a first prompt 61 using a template such as "Please refer to {#Text} and describe the properties of {#Entity} from the perspective of {#Specified property}." The property extraction unit 110 inputs this first prompt 61 to the artificial intelligence 200 and obtains the output from the artificial intelligence 200: "Classification: Rosaceae, Malus genus, Shape: Round, Taste: Sweet". This "Classification: Rosaceae, Malus genus, Shape: Round, Taste: Sweet" is an apple property and is an example of the linked source property information 51.
[0038] In step S106, the property extraction unit 110 generates entity L as source property information 51, which is a pair of the source entity name i and the properties of the source entity. At this time, L will be between 1 and n. In step S107, the property extraction unit 110 stores entity L in the source property database 521.
[0039] The process from step S108 to step S111 is the same as the process from step S104 to step S107, but with the source entity replaced by the destination entity. The process is the same as from step S104 to step S107 as follows.
[0040] In step S108, the property extraction unit 110 obtains linked entity information 22 corresponding to the linked entity name j from the linked entity database 512. The linked entity database 512 is a database that stores information (text, graphs, tables) associated with the entity name j. In step S109, the property extraction unit 110 generates a second prompt 62 to obtain linked property information 52, which is a property representing the linked entity using the perspective 30, from the linked entity information 22. Then, the property extraction unit 110 inputs the second prompt 62 to the artificial intelligence 200 to extract the linked property information 52, which is a property of the linked entity. In step S110, the property extraction unit 110 generates entity K as linked property information 52, which is a pair of linked entity name j and the property of the linked entity. At this time, K is between 1 and m. In step S111, the property extraction unit 110 stores entity K in the linked property database 522.
[0041] <Linking Determination Process: Steps S112 to S114> The linking determination unit 120 uses the source property information 51 and the destination entity information 22 to determine the similarity between the source entity and the destination entity. Based on the similarity determination result, the linking determination unit 120 determines whether data linking is possible. In this embodiment, the linking determination unit 120 uses the source property information 51 obtained from the source entity information 21 and the destination property information 52 obtained from the destination entity information 22 to determine whether data linking is possible. The linking determination unit 120 uses the source property information 51 and the destination property information 52 to determine the similarity between the source entity and the destination entity and determines whether data linking is possible.
[0042] For example, the cooperation determination unit 120 calculates the cosine similarity between the source cooperation property information 51 and the destination cooperation property information 52, and determines the similarity between the source entity and the destination entity based on the cosine similarity. Then, the cooperation determination unit 120 determines whether data cooperation is possible based on the determination result of the similarity between the source cooperation property information 51 and the destination cooperation property information 52. For example, when the cosine similarity is greater than a predetermined threshold value, the cooperation determination unit 120 determines that there is similarity and determines that data cooperation is possible.
[0043] Alternatively, the cooperation determination unit 120 acquires the source cooperation property information 51 and the destination cooperation property information 52, and generates a fourth prompt 64 for outputting whether data cooperation between the source entity and the destination entity is possible from the artificial intelligence 200. Then, the cooperation determination unit 120 may determine whether data cooperation is possible by inputting the fourth prompt 64 to the artificial intelligence 200. Specifically, the cooperation determination unit 120 includes a cooperation candidate acquisition unit 211 and a cooperation determination processing unit 212. It is as follows:
[0044] In step S112, the cooperation candidate acquisition unit 211 of the cooperation determination unit 120 creates a pair of the entity L which is the source cooperation property information 51 and the entity K which is the destination cooperation property information 52. In step S113, the cooperation determination processing unit 212 of the cooperation determination unit 120 determines whether data cooperation between the source entity and the destination entity is possible using the entity L and the entity K. The method for determining whether data cooperation is possible is as described above. In step S114, the cooperation determination unit 120 stores the pair of the entity name i of the entity L and the entity name j of the entity K in the cooperation result database 530.
[0045] In Figure 5, the source entity information 21, "text describing apples," is used to extract the source property information 51, "classification of apples: Rosaceae, Malus genus." Similarly, the destination entity information 22, "text describing peaches," is used to extract the destination property information 52, "classification of peaches: Rosaceae, Prunus genus." Furthermore, the destination entity information 22, "text describing tomatoes," is used to extract the destination property information 52, "classification of tomatoes: Solanaceae, Tomato genus."
[0046] The linkage determination unit 120 then uses the above-mentioned source property information 51 and destination property information 52 to determine the similarity using cosine similarity or artificial intelligence 200, and determines whether or not data linkage is possible.
[0047] When using cosine similarity, the linkage determination unit 120 compares, for example, the vectors of the apple classification "Rosaceae Malus genus" and the tomato classification "Solanaceae Tomato genus," and determines that those exceeding a threshold are similar. If they are determined not to be similar, the linkage determination unit 120 determines that data linkage between apples and tomatoes is not possible. If they are determined to be similar, the linkage determination unit 120 determines that data linkage between apples and tomatoes is possible.
[0048] When using the artificial intelligence 200, the cooperation determination unit 120 creates, for example, the fourth prompt 64 such as "Can apples of the genus Malus in the Rosaceae family and tomatoes of the genus Solanum in the Solanaceae family be data-linked from the perspective of classification?" and queries the artificial intelligence 200. Alternatively, the cooperation determination unit 120 may create the fourth prompt 64 such as "Are the genus Malus in the Rosaceae family and the genus Solanum in the Solanaceae family similar from the perspective of classification?" and query the artificial intelligence 200. If it is determined that they are not similar, the cooperation determination unit 120 determines that data linking between apples and tomatoes is impossible. If it is determined that they are similar, the cooperation determination unit 120 determines that data linking between apples and tomatoes is possible. Also, the fourth prompt 64 may be "Please show the basis for the result of whether apples of the genus Malus in the Rosaceae family and tomatoes of the genus Solanum in the Solanaceae family can be data-linked from the perspective of classification." Alternatively, it may be "Please show the basis for the result of whether the genus Malus in the Rosaceae family and the genus Solanum in the Solanaceae family are similar from the perspective of classification."
[0049] In FIG. 5, as the determination result of whether data linking is possible, it is output that "the 'apple' entity and the 'peach' entity can be linked" and "the 'apple' entity and the 'tomato' entity cannot be linked".
[0050] Note that the entity L and the entity K stored in the source property database 521 and the destination property database 522 can be displayed, for example, according to a user's request or the like.
[0051] ***Other Configurations*** <Modification 1> Figure 7 is a schematic diagram showing an example of the property extraction process according to Modification 1 of this embodiment. In this embodiment, an example was described in which each of the source entity information and the destination entity information is in text format. In Modification 1, each of the source entity information and the destination entity information may be in graph format or table format. As shown in Figure 7, when each of the entity information is in graph format or table format, the properties of the entity are often clear in advance. Specifically, in Figure 7, it is clear that the properties of the apple entity are "Classification: Rosaceae Malus, Shape: Round, Taste: Sweet, Use: Eaten raw, Juice". Therefore, in the prompt to obtain properties, properties similar to the specified property, i.e., the input perspective, are extracted from the list of properties in the entity information.
[0052] <Modification 2> Figure 8 shows an example of the functional configuration of the data linkage device 100 according to Modification 2 of this embodiment. Figure 9 is a flowchart showing an example of the data linkage process according to Modification 2 of this embodiment. Figure 10 is a schematic diagram showing an example of the data linkage process according to Modification 2 of this embodiment.
[0053] In this embodiment, the property extraction unit extracts source property information and destination property information. The linking determination unit then uses the source property information and destination property information to determine whether data linking between the source entity and the destination entity is possible. On the other hand, in modified example 2, the process is as follows.
[0054] As shown in Figures 8 and 9, in the modified example 2, the property extraction unit 110 extracts only the linked property information 51, i.e., entity L, from the linked entity information 21.
[0055] The linking determination unit 120 then acquires the source property information 51 and the destination entity information 22. The linking determination unit 120 generates a third prompt 63 to cause the artificial intelligence 200 to output the result of the similarity determination between the source entity and the destination entity, or whether data linking is possible. By inputting the third prompt 63 to the artificial intelligence 200, the linking determination unit 120 determines the similarity between the source entity and the destination entity, or whether data linking is possible. Specifically, it is as follows.
[0056] In step S112a, the linking candidate acquisition unit 211 of the linking determination unit 120 creates a pair of entity L, which is the source property information 51, and linking destination entity information 22. In step S113a, the linking determination processing unit 212 of the linking determination unit 120 uses the pair of entity L, which is the source property information 51, and linking destination entity information 22 to determine whether data linking between the source entity and the linking destination entity is possible.
[0057] In Figure 10, a pair is created consisting of the source property information 51, "Classification of apple: Rosaceae, Malus genus," and the destination entity information 22, "A sentence describing a peach." Similarly, a pair is created consisting of the source property information 51, "Classification of apple: Rosaceae, Malus genus," and the destination entity information 22, "A sentence describing a tomato." The linking determination processing unit 212 generates a third prompt 63, for example, as follows: The third prompt 63 is generated as, "Apples and tomatoes are similar in terms of classification. Apples belong to the Rosaceae, Malus genus. For the classification of tomatoes, please make your decision based on the following description. Description: 'A sentence describing a tomato'." The linking determination processing unit 212 inputs the third prompt 63 to the artificial intelligence 200 to determine the similarity between the source entity and the destination entity, or whether data linking is possible.
[0058] According to Modification 2, the property extraction process can be reduced, resulting in improved processing efficiency.
[0059] <Modification 3> In this embodiment, the source and destination entities were created separately, but this is not necessary. The source entity information and the destination entity information may be stored in the same entity database. Also, the source property information and the destination property information may each be stored in the same property database.
[0060] Furthermore, in the embodiments described above and those described below, the source and destination entities may be reversed. For example, in Modification 2, the property extraction unit 110 is configured to extract only source property information 51, i.e., entity L, from source entity information 21. However, the property extraction unit 110 may also be configured to extract only destination property information 52, i.e., entity K, from destination entity information 22. The linkage determination unit 120 may then acquire the destination property information 52 and source entity information 21 and determine the similarity between the source entity and the destination entity, or whether data linkage is possible.
[0061] <Modification 4> In this embodiment, the functions of the property extraction unit 110 and the linkage determination unit 120 are implemented in software. As a modification, the functions of the property extraction unit 110 and the linkage determination unit 120 may be implemented in hardware. Specifically, the data linkage device 100 includes an electronic circuit 909 instead of a processor 910.
[0062] Figure 11 shows an example configuration of a data linkage device 100 according to modification 4 of this embodiment. The electronic circuit 909 is a dedicated electronic circuit that realizes the functions of the property extraction unit 110 and the linkage determination unit 120. Specifically, the electronic circuit 909 is a single circuit, a composite circuit, a programmed processor, a parallel programmed processor, a logic IC, a GA, an ASIC, or an FPGA. GA is an abbreviation for Gate Array. ASIC is an abbreviation for Application Specific Integrated Circuit. FPGA is an abbreviation for Field-Programmable Gate Array.
[0063] The functions of the property extraction unit 110 and the cooperation determination unit 120 may be implemented in a single electronic circuit, or they may be implemented in a distributed manner across multiple electronic circuits.
[0064] As another variation, some functions of the property extraction unit 110 and the cooperation determination unit 120 may be implemented by electronic circuits, and the remaining functions may be implemented by software. Alternatively, some or all functions of the property extraction unit 110 and the cooperation determination unit 120 may be implemented by firmware.
[0065] The processor and the electronic circuit are also called processing circuits. In other words, the functions of the property extraction unit 110 and the cooperation determination unit 120 are realized by the processing circuits.
[0066] ***Explanation of the Effects of This Embodiment*** Figure 12 is a schematic diagram illustrating the effects of the data linkage system 500 according to this embodiment. As shown in the left figure of Figure 12, in a data linkage feasibility determination using only cosine similarity, only the result of cosine similarity is known. On the other hand, as shown in the right figure of Figure 12, according to the data linkage feasibility determination according to this embodiment, the properties of the entities to be linked are revealed. As described above, in the data linkage device 100 according to this embodiment, when a person makes the final decision on whether or not to link data, the basis for the decision on whether or not to link data becomes clear. Therefore, there is an effect of reducing the decision burden.
[0067] Furthermore, the data linkage device 100 according to this embodiment extracts properties of the entity to be linked from a certain perspective. The input data is text, i.e., a description, of the entity to be linked. Then, using LLM, properties are extracted from the text of the entity to be linked according to a certain perspective. Therefore, the data linkage device 100 according to this embodiment has the effect of improving the accuracy of data linkage because data linkage can be performed using the specified properties, i.e., perspective, intended by the user. It also has the effect of being able to determine whether or not data linkage is possible even for data in unstructured text format.
[0068] Embodiment 2. This embodiment mainly describes the differences from Embodiment 1 and the additions to Embodiment 1. In this embodiment, components having the same function as in Embodiment 1 are denoted by the same reference numerals, and their descriptions are omitted.
[0069] In Embodiment 1, the viewpoint 30 is input to the data linkage device 100 by the user. In this embodiment, we will describe a mode in which the data linkage device 100 automatically determines the viewpoint 30 when the purpose 80 of data linkage is input by the user.
[0070] ***Description of Configuration*** Figure 13 shows an example of the functional configuration of the data linkage device 100 according to this embodiment. In addition to the functional configuration described in Embodiment 1, the data linkage device 100 according to this embodiment includes a viewpoint determination unit 160. The other configurations are the same as in Embodiment 1.
[0071] ***Explanation of Operation*** Figure 14 is a flowchart showing an example of the data linkage process according to this embodiment. Figure 15 is a schematic diagram showing an example of the data linkage process according to this embodiment. In Figure 14, the process corresponding to step S101 described in Embodiment 1 becomes steps S101a and S101b.
[0072] In step S101a, the perspective determination unit 160 acquires the purpose of data linkage 80 and information from at least one of the source entity information 21 and the destination entity information 22. The perspective determination unit 160 acquires the purpose of data linkage 80 and information from at least one of the source entity information 21 and the destination entity information 22 via the input interface 930.
[0073] Let's explain this in detail using Figure 15. In Figure 15, the "Book XX" entity is either the source or destination entity for the data linkage. Here, we will assume it is the source entity. The user inputs "a description of Book XX" into the perspective determination unit 160. The user also inputs two objectives 80 for the data linkage into the perspective determination unit 160: "I want to find similar books" and "I want to find books that are identical in terms of copyright." The user inputs the above information into the data linkage device 100 via the input interface 930.
[0074] In step S101b, the viewpoint determination unit 160 generates a fifth prompt 65 to cause the artificial intelligence 200 to output the viewpoint 30 to be used when performing data linkage for the acquired data linkage purpose 80. The viewpoint determination unit 160 acquires the viewpoint 30 by inputting the fifth prompt 65 to the artificial intelligence 200.
[0075] Let's explain this in detail using Figure 15. The perspective determination unit 160 generates a fifth prompt 65, for example, "Please tell me the perspectives to pay attention to when linking data for the input purpose," to cause the artificial intelligence 200 to output the perspective 30. The perspective determination unit 160 inputs this fifth prompt 65 to the artificial intelligence 200. In Figure 15, for the purpose of "I want to find similar books," the perspectives "author, field" are output. Also, for the purpose of "I want to find books that are identical in terms of copyright," the perspective "DOI" is output. DOI is an abbreviation for Digital Object Identifier.
[0076] The processing in steps S102 to S114 is the same as in Figure 4. The processing in steps S104 to S107 is the process of extracting the source property information 51 from the viewpoint 30 obtained in step S101b and the source entity information 21. The processing in steps S108 to S111 is the process of extracting the destination property information 52 from the viewpoint 30 obtained in step S101b and the destination entity information 22.
[0077] ***Explanation of the Effects of This Embodiment*** As described above, the data linkage device 100 according to this embodiment can automatically acquire perspectives that match the purpose of data linkage. Therefore, perspectives can be acquired even without knowledge of the input data. Furthermore, perspectives and properties can be acquired even without knowing what perspectives represent the properties in the input data.
[0078] Embodiment 3. This embodiment mainly describes the differences from Embodiments 1 and 2, and the additions made to Embodiments 1 and 2. In this embodiment, components having the same function as those in Embodiments 1 and 2 are denoted by the same reference numerals, and their descriptions are omitted.
[0079] In Embodiment 1, the source entity database 511 has source entity information 21 stored in advance. The destination entity database 512 has destination entity information 22 stored in advance. The user inputs the source entity name and destination entity name into the data linkage device 100. In this embodiment, when the user inputs the purpose 80 of the data linkage, the data linkage device 100 will be described in which it automatically identifies the source or destination entity information from the input data.
[0080] ***Description of Configuration*** Figure 16 shows an example of the functional configuration of the data linkage device 100 according to this embodiment. In addition to the functional configuration described in Embodiment 1, the data linkage device 100 according to this embodiment includes an entity identification unit 170. The other configurations are the same as in Embodiment 1.
[0081] ***Explanation of Operation*** Figure 17 is a flowchart showing an example of the data linkage process according to this embodiment. Figure 18 is a schematic diagram showing an example of the data linkage process according to this embodiment. In Figure 17, the process corresponding to steps S102 and S103 described in Embodiment 1 consists of step S102a, steps S201 to S203, and steps S301 to S303.
[0082] The entity identification unit 170 acquires the purpose 80 of data linkage and the input data 90 entered for use in data linkage. The entity identification unit 170 generates a sixth prompt 66 for generating information associated with the entity to be used when linking data for the acquired purpose from the input data 90. The entity identification unit 170 acquires information associated with the entity by inputting the sixth prompt 66 to the artificial intelligence 200. The property extraction unit 110 uses the information associated with the entity as either the source entity information 21 or the destination entity information 22. Specifically, it is as follows.
[0083] In step S102a, the entity identification unit 170 obtains the purpose of data linkage 80 from the user via the input interface 930.
[0084] In step S201, the entity identification unit 170 generates information from the input data 90 that is associated with the source entity used when linking data with the input objective 80. Here, it is assumed that the input data 90 is stored in advance in the source entity database 511. In step S301, the entity identification unit 170 generates information from the input data 90 that is associated with the destination entity used when linking data with the input objective 80. Here, it is assumed that the input data 90 is stored in advance in the destination entity database 512. Specifically, it is as follows.
[0085] The entity identification unit 170 generates a sixth prompt 66 for generating information associated with the source entity used when linking data for the input purpose 80, from the input data 90. The entity identification unit 170 obtains information associated with the source entity by inputting the sixth prompt 66 to the artificial intelligence 200. Information associated with the source entity is, for example, a text, graph, or table that describes the source entity. Information associated with the source entity is the source entity information 21. In step S202, the entity identification unit 170 outputs the source entity information 21 to the source entity database 511. The source entity information 21 consists of an entity name i and information associated with it. In step S203, the entity identification unit 170 collects all the output source entity names, i.e., entity names i, into an entity name set I.
[0086] Furthermore, the entity identification unit 170 generates a sixth prompt 66 for generating information associated with the linked entity to be used when linking data for the input purpose 80, from the input data 90. The entity identification unit 170 obtains information associated with the linked entity by inputting the sixth prompt 66 to the artificial intelligence 200. The information associated with the linked entity is, for example, a text, graph, or table that describes the linked entity. The information associated with the linked entity is the linked entity information 22. In step S302, the entity identification unit 170 outputs the linked entity information 22 to the linked entity database 512. The linked entity information 22 consists of an entity name j and information associated with it. In step S303, the entity identification unit 170 collects all the outputted linked entity names, i.e., entity names j, into an entity name set J.
[0087] Let's explain this in detail using Figure 18. The data that the user uses to achieve their objective in data linkage is called input data 90. Input data 90 can be text, graphs, or tables. In Figure 18, the user inputs the objective 80 of data linkage, "I want to find people with similar backgrounds," and the above-mentioned input data 90 into the entity identification unit 170. The user inputs the above information into the data linkage device 100 via the input interface 930.
[0088] The entity identification unit 170 generates a sixth prompt 66 to generate information associated with the source entity used when linking data for the input purpose 80, from the input data 90. For example, the sixth prompt 66 is generated as, "Identify the entity to be focused on when linking data for the input purpose, and create a description of it." The entity identification unit 170 inputs this sixth prompt 66 to the artificial intelligence 200. The artificial intelligence 200 outputs information associated with "Mr. A" as the source entity. Here, the artificial intelligence 200 outputs "a description of Mr. A," that is, source entity information 21 with "Mr. A" as the source entity.
[0089] In the above explanation, it is assumed that the input data 90 is stored in advance in the entity databases of the source and destination. However, it may also be text, graphs, or tables entered by the user from sources other than the entity databases of the source and destination. The entity identification unit 170 may also obtain information associated with entities from the input data 90 obtained from sources other than the entity databases of the source and destination. In this case, the property extraction unit 110 uses the information associated with the entity as either source entity information or destination entity information.
[0090] ***Explanation of the Effects of This Embodiment*** As described above, the data linkage device 100 according to this embodiment can automatically acquire entities that match the purpose of data linkage. Therefore, entities can be acquired even without knowledge of the input data. Furthermore, entities can be acquired even without knowing what entities are in the input data. Therefore, the data linkage device 100 according to this embodiment can extract properties that are not explicitly stated in the input data. In other words, possessed knowledge can be made into properties. In addition, the number of usable perspectives increases, and the accuracy of data linkage improves. In the example in Figure 18, the property that person A possesses knowledge about generated AI is not explicitly stated in the input data. However, the description of the person A entity explicitly states, "Has knowledge about generated AI."
[0091] In embodiments 1 to 3 described above, each part of the data linkage device was described as an independent functional block. However, the configuration of the data linkage device does not have to be as described in the embodiments above. The functional blocks of the data linkage device can be configured in any way as long as they can realize the functions described in the embodiments above. Also, the data linkage device does not have to be a single device, but a system composed of multiple devices. Furthermore, multiple parts of embodiments 1 to 3 may be combined and implemented. Alternatively, only one part of these embodiments may be implemented. In addition, these embodiments may be combined and implemented in any way, either as a whole or in part. That is to say, in embodiments 1 to 3, it is possible to freely combine each embodiment, modify any component of each embodiment, or omit any component in each embodiment.
[0092] The embodiments described above are essentially preferred examples and are not intended to limit the scope of the Disclosure, the scope of the Applications of the Disclosure, or the scope of Uses of the Disclosure. The embodiments described above can be modified in various ways as needed. For example, the procedures described using flowcharts or sequence diagrams may be modified as appropriate.
[0093] The various aspects of this disclosure are summarized below as an appendix.
[0094] (Note 1) A data linkage device comprising: a property extraction unit that obtains source entity information, which is information associated with the source entity in data linkage, and a viewpoint representing the source entity; generates a first prompt for generating source property information, which is a property representing the source entity using the viewpoint, from the source entity information; and inputs the first prompt to artificial intelligence to extract the source property information from the source entity information; and a linkage determination unit that determines whether data linkage between the source entity and the linked entity is possible, using the source property information and linked entity information, which is information associated with the linked entity in data linkage. (Note 2) The property extraction unit acquires the linked entity information, generates a second prompt for generating linked property information which is a property representing the linked entity using the viewpoint, from the linked entity information, and extracts the linked property information from the linked entity information by inputting the second prompt to the artificial intelligence. The linkage determination unit uses the linked property information and the linked property information to determine whether data linkage between the linked entity and the linked entity is possible, as described in Note 1. (Note 3) The linkage determination unit calculates the cosine similarity between the linked property information and the linked property information, and determines whether data linkage between the linked entity and the linked entity is possible based on the cosine similarity, as described in Note 2. (Note 4) The data linkage device according to Note 1, wherein the linkage determination unit generates a third prompt to cause the artificial intelligence to output whether or not data linkage between the linkage source entity and the linkage destination entity is possible, using the linkage source property information and the linkage destination entity information, and determines whether or not data linkage between the linkage source entity and the linkage destination entity is possible by inputting the third prompt to the artificial intelligence.(Note 5) The data linkage device according to Note 2, wherein the linkage determination unit generates a fourth prompt to cause the artificial intelligence to output whether or not data linkage between the linkage source entity and the linkage destination entity is possible, using the linkage source property information and the linkage destination property information, and determines whether or not data linkage between the linkage source entity and the linkage destination entity is possible by inputting the fourth prompt to the artificial intelligence. (Note 6) The data linkage device according to any one of Notes 1 to 5, further comprising a viewpoint determination unit that acquires the purpose of the data linkage and information from at least one of the linkage source entity information and the linkage destination entity information, generates a fifth prompt to cause the artificial intelligence to output a viewpoint to be used when performing the data linkage for the purpose, and acquires the viewpoint by inputting the fifth prompt to the artificial intelligence. (Note 7) The data linkage device includes an entity identification unit that acquires the purpose of the data linkage and input data entered for use in the data linkage, generates a sixth prompt for generating information associated with entities used when performing the data linkage for the purpose from the input data, and acquires information associated with entities by inputting the sixth prompt to artificial intelligence, and the property extraction unit uses the information associated with entities as either the source entity information or the destination entity information, as described in any one of Notes 1 to 6. (Note 8) The data linkage device according to any one of Notes 1 to 7, wherein each of the source entity information and the destination entity information is in document format, graph format or table format.(Note 9) A data linkage system comprising artificial intelligence and a data linkage device for determining whether data linkage between a source entity and a destination entity is possible, comprising: a property extraction unit that acquires source entity information, which is information associated with the source entity, and a viewpoint representing the source entity, generates a first prompt for generating source property information, which is a property representing the source entity using the viewpoint, from the source entity information, and inputs the first prompt to the artificial intelligence to extract the source property information from the source entity information; and a linkage determination unit that determines whether data linkage between the source entity and the destination entity is possible using the source property information and destination entity information, which is information associated with the destination entity. (Note 10) A data linkage method comprising: a computer obtaining source entity information, which is information associated with the source entity in data linkage, and a viewpoint representing the source entity; generating a first prompt for generating source property information, which is a property representing the source entity, from the source entity information using the viewpoint; inputting the first prompt to artificial intelligence to extract the source property information from the source entity information; and the computer determining whether data linkage between the source entity and the destination entity is possible, using the source property information and destination entity information, which is information associated with the destination entity in data linkage.(Note 11) A data linkage program that causes a computer to execute a property extraction process that extracts the source property information from the source entity information by inputting the first prompt to an artificial intelligence, which is a property that represents the source entity, which is a property that represents the source entity, which is a property that represents the source entity, which is a property that represents the source entity, which is a property that extracts the source property information from the source entity information, and which determines whether or not data linkage is possible between the source entity and the destination entity using the source property information and destination entity information that is a property that represents the source entity in data linkage.
[0095] 30 Perspective, 21 Source entity information, 22 Destination entity information, 51 Source property information, 52 Destination property information, 61 First prompt, 62 Second prompt, 63 Third prompt, 64 Fourth prompt, 65 Fifth prompt, 66 Sixth prompt, 80 Purpose, 90 Input data, 100 Data linkage device, 110 Property extraction unit, 120 Linkage determination unit, 211 Linkage candidate acquisition unit, 212 Linkage determination processing unit, 150 Storage unit, 160 Perspective determination unit, 170 Entity identification unit, 511 Source entity database, 512 Destination entity database, 521 Source property database, 522 Destination property database, 530 Linkage result database, 200 Artificial intelligence, 500 Data linkage system, 909 Electronic circuit, 910 Processor, 921 Memory, 922 Auxiliary storage device, 930 Input interface, 940 Output interface, 950 Communication device.
Claims
1. A data linkage device comprising: a property extraction unit that obtains source entity information, which is information associated with the source entity in data linkage, and a perspective representing the source entity; generates a first prompt for generating source property information, which is a property representing the source entity, from the source entity information using the perspective; and inputs the first prompt to artificial intelligence to extract the source property information from the source entity information; and a linkage determination unit that determines whether data linkage between the source entity and the linked entity is possible, using the source property information and linked entity information, which is information associated with the linked entity in data linkage.
2. The data linkage device according to claim 1, wherein the property extraction unit acquires the linked entity information, generates a second prompt for generating linked property information which is a property representing the linked entity using the viewpoint, and extracts the linked property information from the linked entity information by inputting the second prompt to the artificial intelligence; and the linkage determination unit determines whether data linkage between the linked entity and the linked entity is possible using the linked property information and the linked property information.
3. The data linkage device according to claim 2, wherein the linkage determination unit calculates the cosine similarity between the source property information and the destination property information, and determines whether or not data linkage between the source entity and the destination entity is possible based on the cosine similarity.
4. The data linkage device according to claim 1, wherein the linkage determination unit generates a third prompt to cause the artificial intelligence to output whether or not data linkage between the linkage source entity and the linkage destination entity is possible, using the linkage source property information and the linkage destination entity information, and determines whether or not data linkage between the linkage source entity and the linkage destination entity is possible by inputting the third prompt to the artificial intelligence.
5. The data linkage device according to claim 2, wherein the linkage determination unit generates a fourth prompt to cause the artificial intelligence to output whether or not data linkage between the linkage source entity and the linkage destination entity is possible, using the linkage source property information and the linkage destination property information, and determines whether or not data linkage between the linkage source entity and the linkage destination entity is possible by inputting the fourth prompt to the artificial intelligence.
6. The data linkage device according to any one of claims 1 to 5, further comprising: a data linkage device that acquires the purpose of the data linkage and information of at least one of the source entity information and the destination entity information; a viewpoint determination unit that generates a fifth prompt to cause the artificial intelligence to output a viewpoint to be used when performing the data linkage for the purpose; and a viewpoint determination unit that acquires the viewpoint by inputting the fifth prompt to the artificial intelligence.
7. The data linkage device comprises an entity identification unit that acquires the purpose of the data linkage and input data entered for use in the data linkage, generates a sixth prompt for generating information associated with an entity used when performing the data linkage for the purpose from the input data, and acquires information associated with the entity by inputting the sixth prompt to artificial intelligence, and the property extraction unit uses the information associated with the entity as either the source entity information or the destination entity information, according to any one of claims 1 to 6.
8. The data linkage device according to any one of claims 1 to 7, wherein each of the source entity information and the destination entity information is in document format, graph format, or table format.
9. A data linkage system comprising artificial intelligence and a data linkage device for determining whether data linkage between a source entity and a destination entity is possible, comprising: a property extraction unit that acquires source entity information, which is information associated with the source entity, and a viewpoint representing the source entity; generates a first prompt for generating source property information, which is a property representing the source entity, from the source entity information using the viewpoint; and inputs the first prompt to the artificial intelligence to extract the source property information from the source entity information; and a linkage determination unit that determines whether data linkage between the source entity and the destination entity is possible, using the source property information and destination entity information, which is information associated with the destination entity.
10. A data linkage method comprising: a computer obtaining source entity information, which is information associated with the source entity in data linkage, and a viewpoint representing the source entity; generating a first prompt for generating source property information, which is a property representing the source entity, from the source entity information using the viewpoint; inputting the first prompt to artificial intelligence to extract the source property information from the source entity information; and the computer determining whether data linkage between the source entity and the destination entity is possible, using the source property information and destination entity information, which is information associated with the destination entity in data linkage.
11. A data linkage program that causes a computer to perform the following: a property extraction process that extracts the source property information from the source entity information by inputting the first prompt to an artificial intelligence, by obtaining source entity information, which is information associated with the source entity in data linkage, and a viewpoint representing the source entity; a linkage determination process that determines whether data linkage between the source entity and the linked entity is possible, using the source property information and linked entity information, which is information associated with the linked entity in data linkage.