Information processing device

The information processing apparatus addresses the challenge of limited information expansion in knowledge graphs by linking and associating entities between graphs, enabling efficient and accurate data augmentation.

JP2026182094APending Publication Date: 2026-11-19NEC CORP
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
JP2025078023
Authority / Receiving Office
JP · JP
Patent Type
Applications
Current Assignee / Owner
Filing Date
2025-05-08
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing knowledge graphs struggle with limited information expansion capabilities, as they do not easily accommodate additional data, affecting their utility across various fields.

Method used

An information processing apparatus that links and associates entities between first and second knowledge graphs using a first linking unit, extracts candidate entities, and calculates relevance to enhance the information content by connecting unlinked entities.

Benefits of technology

Facilitates easy expansion of knowledge graph information by linking entities across different languages and notations, enhancing the accuracy and quantity of data within the graphs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 2026182094000001_ABST
    Figure 2026182094000001_ABST
Patent Text Reader

Abstract

The inability to easily increase the amount of information in the knowledge graph. [Solution] The information processing device of the present disclosure comprises: a first linking unit that associates a second entity with a first entity based on a first knowledge graph consisting of a first entity and a first relation in a first form of notation, a second form of notation of the first knowledge graph, and a second knowledge graph consisting of a second entity and a second relation in a second form of notation; a candidate extraction unit that extracts candidate second entities from the second knowledge graph that are candidates for second entities to be associated with a non-linked first entity based on a linked first entity linked to a non-linked first entity by a first relation; and a second linking unit that associates the extracted candidate second entities with a non-linked first entity.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present disclosure relates to an information processing apparatus.

Background Art

[0002] As described in Patent Document 1, generating knowledge graphs that represent the relevance of diseases, treatments, drugs, and the like has been performed. Specifically, in Patent Document 1, entities are acquired from documents related to terms input by a physician, and the entities are associated with each other to generate a knowledge graph.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] However, Patent Document 1 does not describe the connection between existing knowledge graphs. For this reason, there is a problem that the amount of information in the knowledge graph cannot be easily increased. Further, not limited to knowledge graphs in the medical field, there is a similar problem that the amount of information cannot be easily increased for knowledge graphs in all fields.

[0005] Accordingly, one of the objects of the present disclosure is to solve the above-described problem that the amount of information in the knowledge graph cannot be easily increased.

Means for Solving the Problems

[0006] An information processing apparatus according to one aspect of the present disclosure is A first linking unit that associates the second entity with the first entity based on a first knowledge graph consisting of first entities in a first form of notation and first relations connecting the first entities, a second form of notation of the first knowledge graph, and a second knowledge graph consisting of second entities in a second form of notation and second relations connecting the second entities, A candidate extraction unit extracts candidate second entities from the second knowledge graph that are candidates for the second entity to be associated with the non-linked first entity, based on a linked first entity, which is another first entity connected by the first relation to a predetermined non-linked first entity, which is a first entity in the first knowledge graph that is not associated with the second entity; A second linking unit that associates the extracted candidate second entity with the first non-linked entity, Equipped with, This is the structure it takes. Furthermore, the information processing method, which is one form of this disclosure, Information processing device, Based on a first knowledge graph consisting of first entities in first form notation and first relations connecting said first entities, a second form notation of said first knowledge graph, and a second knowledge graph consisting of second entities in second form notation and second relations connecting said second entities, the second entities are associated with the first entities. Based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and another first entity, which is a linked first entity, that is connected by the first relation, a candidate second entity that can be associated with the non-linked first entity is extracted from the second knowledge graph. The extracted candidate second entity is associated with the first non-linked entity. This is the structure it takes. Furthermore, one form of this disclosure is a program, In an information processing device, Based on a first knowledge graph consisting of first entities in first form notation and first relations connecting said first entities, a second form notation of said first knowledge graph, and a second knowledge graph consisting of second entities in second form notation and second relations connecting said second entities, the second entities are associated with the first entities. Based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and another first entity, which is a linked first entity, that is connected by the first relation, a candidate second entity that can be associated with the non-linked first entity is extracted from the second knowledge graph. The extracted candidate second entity is associated with the first non-linked entity. To execute the process This is the structure it takes. [Effects of the Invention]

[0007] This disclosure, configured as described above, makes it easy to increase the amount of information in the knowledge graph. [Brief explanation of the drawing]

[0008] [Figure 1] This block diagram shows an example of the configuration of the information processing device in this disclosure. [Figure 2] This flowchart shows an example of processing operations by the information processing device described in this disclosure. [Figure 3] This figure shows an example of processing by the information processing device described in this disclosure. [Figure 4] This figure shows an example of processing by the information processing device described in this disclosure. [Figure 5]This is a diagram illustrating an example of processing by the information processing apparatus in the present disclosure. [Figure 6] This is a diagram illustrating an example of processing by the information processing apparatus in the present disclosure. [Figure 7] This is a diagram illustrating an example of processing by the information processing apparatus in the present disclosure. [Figure 8] This is a block diagram illustrating an example of the hardware configuration of the information processing apparatus in the present disclosure. [Figure 9] This is a block diagram illustrating an example of the configuration of the information processing apparatus in the present disclosure. [Figure 10] This is a flowchart illustrating an example of processing operations by the information processing apparatus in the present disclosure. [Mode for Carrying Out the Invention]

[0009] <First Embodiment> A first embodiment of the present disclosure will be described with reference to the drawings. Note that the drawings may be relevant to any of the embodiments.

[0010] As an example, the information processing apparatus 10 of the present disclosure is used for mutually associating and linking knowledge graphs representing relevance in diseases, treatments, pharmaceuticals, and the like. Specifically, in the present embodiment, description will be given by taking as an example a case where a Japanese knowledge graph is extracted from an electronic medical record 30 storing medical-related information and the like of a predetermined person, translated into English, and the English knowledge graph in the external database 40 is linked to the aforementioned knowledge graph.

[0011] However, the knowledge graphs targeted for linking in the present disclosure are not limited to knowledge graphs related to medical information in the medical field, and are also applicable to knowledge graphs related to any information in any field. In addition, the language and notation of the knowledge graphs targeted for linking in the present disclosure may be any language and any form of notation.

[0012] Hereinafter, an example of the configuration and operation of the information processing apparatus 10 in the present embodiment will be described. The information processing apparatus 10 is configured by one or a plurality of information processing apparatuses each including an arithmetic device and a storage device. As shown in Figure 1, the information processing apparatus 10 includes an extraction unit 11, a translation unit 12, an entity linking unit 13, an unlinked entity extraction unit 14, a search origin entity extraction unit 15, a linking candidate extraction unit 16, a relevance calculation unit 17, and an unlinked entity linking unit 18. The respective functions of the extraction unit 11, translation unit 12, entity linking unit 13, unlinked entity extraction unit 14, search origin entity extraction unit 15, linking candidate extraction unit 16, relevance calculation unit 17, and unlinked entity linking unit 18 can be realized by the arithmetic device executing a program stored in the storage device for realizing each of the functions. The information processing apparatus 10 further includes a search rule storage unit 21 and a graph data storage unit 22, which are implemented by the storage device. An electronic medical record 30 and an external database 40 are connected to the information processing apparatus 10.

[0013] The electronic medical record 30 is a data file in which medical information based on a doctor's diagnosis of a predetermined person is stored. In the present embodiment, it is assumed that medical information such as diseases, treatments, and drugs is written in the electronic medical record 30 as text mainly in Japanese (the first language (notation in the first format)).

[0014] The external database 40 is a database of medical information and stores a knowledge graph related to medical information. The knowledge graph is a graphical representation of named entities such as diseases, treatments, and drugs, linked to their relationships (relationships). In this embodiment, the knowledge graph in the external database 40 is written in English (second language (second form of notation)) and will be referred to as the second knowledge graph below, with entities within the second knowledge graph being referred to as second entities and relationships as second relations. Examples of external databases 40 include ICD (International Classification of Diseases), ATC (Anatomical Therapeutic Chemical Classification System), MeSH (Medical Subject Headings), SNOMED CT (Systematized Nomenclature of Medicine--Clinical Terms), UMLS (Unified Medical Language System), and ChEMBL.

[0015] The extraction unit 11 (conversion unit) recognizes and extracts named entities from the Japanese text in the electronic medical record 30, and further extracts the relationships between the extracted named entities (step S1 in Figure 2). Hereafter, named entities will be referred to as "entities," relationships as "relations," and a graph consisting of entities and relations connecting entities will be referred to as a knowledge graph. For example, as shown in Figure 3, the extraction unit 11 extracts a Japanese knowledge graph in which entity E1 "Disease A" and entity E2 "Surgery X" are connected by relation R "Treatment." In this embodiment, the knowledge graph extracted from the electronic medical record 30 will be referred to as the first knowledge graph, entities within the first knowledge graph will be referred to as the first entities, and relations will be referred to as the first relations.

[0016] Here, the extraction unit 11 applies, for example, a Named Entity Recognition (NER) technique to extract entities such as diseases, symptoms, and medical procedures that appear in the text. Furthermore, the extraction unit 11 identifies the relationships between the extracted entities from the text using, for example, a Relationship Extraction (RE) technique. For example, if the text of the electronic medical record 30 contains a description such as "~ performed for ~", the extraction unit 11 detects a relationship between a pair of entities that have the relationship that a certain procedure was performed for a certain disease.

[0017] As described above, the translation unit 12 (conversion unit) translates the first entities E1, E2 and the first relation R, which constitute the first knowledge graph extracted from the electronic medical record 30 in Japanese, into English (step S2 in Figure 2). The translation unit 12 then stores the translated English terms in the first knowledge graph. For example, as shown in Figure 3, the translation unit 12 stores the first entity E1 "Disease Name A" in both Japanese and English, "DISEASE A", the first entity E2 "Surgery X" in both Japanese and English, "Surgery X", and the first relation R "Treatment" in both Japanese and English.

[0018] The entity linking unit 13 (first linking unit) compares and links the first knowledge graph extracted and translated from the electronic medical record 30 with the second knowledge graph in the database 40, as described above (step S3 in Figure 2). Specifically, the entity linking unit 13 performs a string matching search for the Japanese or English notation held by the first entities E1 and E2 of the first knowledge graph against the second entity of the second knowledge graph in the database 40. Then, the entity linking unit 13 links the second entity of the second knowledge graph in the external database 40, which matches either the Japanese or English notation of the first entities E1 and E2 of the first knowledge graph, to the first entities E1 and E2 of the first knowledge graph.

[0019] Specifically, in the example in Figure 4, the first entity E1 "Disease A" in the first knowledge graph matches the second entity DE1 "A01(ICD),jp:Disease A,en:Disease A" in the external database 40 in either Japanese or English. Therefore, the first entity E1 "Disease A" in the first knowledge graph is linked to the second entity DE1 "A01(ICD)" in the external database 40. At this time, the first entity E1 "Disease A" in the first knowledge graph is assigned the code "A01(ICD)" held in the second entity DE1 in the external database 40 that is being linked. On the other hand, the first entity E2 "Operation xx" in the first knowledge graph does not match the entity in the external database 40 in either Japanese or English, so no linking is performed. In other words, the English name "operation xx" for the first entity E2 in the first knowledge graph and the English name "OPERATION X" for the second entity DE2 in the external database 40 do not exactly match, so linking does not occur.

[0020] The non-linked entity extraction unit 14 (candidate extraction unit) extracts first entities that are not linked to the second entity from the first knowledge graph, which is linked to the second knowledge graph as described above, as non-linked entities (step S4 in Figure 2). In the example in Figure 5, the first entity E2 "operation xx" in the first knowledge graph is extracted as a non-linked entity because it is not linked to the second entity.

[0021] The search starting entity extraction unit 15 (candidate extraction unit) extracts a search starting entity from the first knowledge graph, which is linked to the second knowledge graph as described above, to serve as the starting point for searching for linking candidates for unlinked entities in the second knowledge graph within the external database 40 (step S5 in Figure 2). Specifically, the search starting entity extraction unit 15 extracts the other first entity linked to the unlinked entity in the first knowledge graph by a first relation as the search starting entity. In the example in Figure 5, the first entity E1 "DISEARE A / A01(ICD)", which is linked from the first entity E2 "operation xx" in the first knowledge graph by a first relation "treatment", is extracted as the search starting entity.

[0022] The linking candidate extraction unit 16 (candidate extraction unit) extracts linking candidates (candidate second entities) that can be linked to non-linked entities from the second knowledge graph in the external database 40, starting from the search origin entity described above (step S6 in Figure 2). Specifically, the linking candidate extraction unit 16 first determines a search rule based on the search origin entity and the first relation linked to the search origin entity. At this time, the search rule is predefined according to the entity and the relation linked to the entity and is stored in the search rule storage unit 21. As an example, if the search rule storage unit 21 has a search rule associated with "[search origin entity]-[rule 1]" as "[entity corresponding to search origin entity]-[rule 2]-[rule 3]-[linking candidate]", then the unit determines such a search rule. In the example in Figure 5, the search rule corresponding to the first entity E1, which is the search origin entity, and the first relation R linked to it is selected and determined from the search rule storage unit 21. As a result, as shown in Figure 5, a search rule is determined in which entities located at the end of a chain of relationships and entities linked from the second entity DE1, which corresponds to the first entity E1, the search starting entity, are considered linking candidates.

[0023] Then, the linking candidate extraction unit 16 extracts linking candidates from the second knowledge graph, starting from the search origin entity described above, according to the determined search rule. In the example in Figure 6, in the second knowledge graph in the external database 40, the second entity DE2 "123(SNOMED-CT):OPERATION X" is extracted as a linking candidate by searching by traversing the second relation and second entity in the same way as the determined search rule, starting from the second entity DE1 corresponding to the search origin entity.

[0024] The relevance calculation unit 17 (second linking unit) calculates the degree of relevance between the extracted linking candidates and the electronic medical record 30 from which the first knowledge graph was extracted (step S7 in Figure 2). For example, the relevance calculation unit 17 calculates the similarity between the linking candidate and the link entity, which is the first entity extracted from the text within the same electronic medical record 30 from which the first knowledge graph was extracted and linked with the second entity, and uses the value based on this similarity as the relevance score. In this case, as shown in Figure 7, the relevance calculation unit 17 first calculates word vectors for the extracted linking candidates and the link entities in the electronic medical record 30 using a natural language model (Word2Vec, BERT, etc.) pre-trained with medical terminology. Then, for each linking candidate vector, the relevance calculation unit 17 calculates the similarity between the linking candidate vector and each link entity vector, and uses the average of these similarities as the relevance score of the linking candidate.

[0025] Here, as shown in Figure 7, multiple linking candidates DE2, DE3, and DE4 are extracted, and it is assumed that the first entities E3 and E4 are link entities within the same electronic medical record 30. In this case, for linking candidate DE2, the similarity between the vector of linking candidate DE2 and the vector of link entity E3 is calculated, and the similarity between the vector of linking candidate DE2 and the vector of link entity E4 is calculated, and the relevance is calculated by averaging these similarities. Similarly, for linking candidates DE3 and DE4, the relevance is calculated by averaging the similarities with each link entity E3 and E4, respectively. In the example in Figure 7, it is assumed that the relevance for linking candidate DE2 is calculated to be 0.8, for linking candidate DE3 to be 0.3, and for linking candidate DE4 to be 0.2.

[0026] The non-linked entity linking unit 18 (second linking unit) links each linking candidate to a non-linked entity in the first knowledge graph based on the relevance of each linking candidate calculated as described above (step S8 in Figure 2). For example, the non-linked entity linking unit 18 performs threshold processing on the relevance of the linking candidates and links the linking candidates that clear the threshold processing to the non-linked entities. In this case, if there are multiple linking candidates, the linking candidate with the highest relevance is selected, threshold processing is performed on that linking candidate, and the linking candidate that clears the threshold processing is linked to the non-linked entity. The threshold processing is cleared when the relevance value of a linking candidate is greater than a preset threshold.

[0027] Here, we will explain the case where the correlation between the three linking candidates DE2, DE3, and DE4, as shown in Figure 7, is calculated. The threshold is set to 0.5. In this case, first, linking candidate DE2, which has the highest correlation, is redefined as a single linking candidate. Then, it is determined whether the correlation of the single linking candidate DE2 (0.8) is greater than the threshold of 0.5. If it is greater, the thresholding process is cleared.

[0028] As shown in Figure 7, the non-linked entity linking unit 18 associates the second entity DE2 of the second knowledge graph, which is a linking candidate DE2 whose relevance meets the threshold, with the non-linked entity E2 of the first knowledge graph and links them. As a result, the first entity E2 of the first knowledge graph, "Surgery X / operation xx", which is a non-linked entity, is assigned the code "123 (SNOMED-CT)" held in the second entity DE2 in the external database 40 to be linked. Then, as described above, the non-linked entity linking unit 18 stores the first knowledge graph, in which the second entities of the second knowledge graph have been linked to each first entity, in the graph data storage unit 22.

[0029] As described above, this disclosure first converts a first knowledge graph written in Japanese (the first language) into English (the second language), and then links the entities of the first knowledge graph with the entities of an existing second knowledge graph written in English. In addition, even for unlinked entities of the first knowledge graph that could not be linked, the second knowledge graph can be searched starting from the linked first entities to extract second entities that are candidates for linking, and then linked. This enables linking between knowledge graphs and facilitates an increase in the amount of information in the knowledge graphs.

[0030] Furthermore, this disclosure enables more accurate linking between knowledge graphs by utilizing relevance, such as similarity to entities in the first knowledge graph, when determining which entities to actually link from the linking candidates.

[0031] <Second Embodiment> Next, a second embodiment of the present disclosure will be described with reference to the drawings. The drawings may be relevant to either embodiment.

[0032] The information processing device 10 in this embodiment has the same configuration as the embodiment 1 described above. In addition, the information processing device 10 has the following configuration. The following mainly describes the configuration that differs from the above.

[0033] As described above, the linking candidate extraction unit 16 (candidate extraction unit) of the information processing device 10 in this embodiment is not necessarily limited to extracting linking candidates using search rules. For example, the linking candidate extraction unit 16 may extract all or part of other second entities in the second knowledge graph that can be traced from the second entity corresponding to the search starting entity as linking candidates. Thus, the linking candidate extraction unit 16 may extract linking candidates from the second knowledge graph by any method.

[0034] In this embodiment, the relevance calculation unit 17 of the information processing device 10 may calculate the relevance using any method without using the similarity of the linking candidate and the link entity based on a vector representation. For example, the relevance calculation unit 17 may calculate the relevance based on a dictionary such as a thesaurus.

[0035] Furthermore, the relevance calculation unit 17 and the non-linked entity linking unit 18 (second linking unit) of the information processing device 10 in this embodiment may link the linking candidate to the non-linked entity of the first knowledge graph without necessarily calculating the relevance between the linking candidate and the linked entity. For example, if there is only one extracted linking candidate, that linking candidate may be linked to the non-linked entity of the first knowledge graph.

[0036] <Third Embodiment> Next, a third embodiment of this disclosure will be described with reference to the drawings. The drawings may be relevant to any of the embodiments.

[0037] The information processing device 10 in this embodiment has the same configuration as in embodiments 1 and 2 described above. Furthermore, this embodiment is not limited to linking knowledge graphs written in Japanese and English, but is applicable to linking knowledge graphs between any languages. In addition, this embodiment is not limited to cases where the knowledge graph is written in language, but is also applicable when linking knowledge graphs written in mutually different notations such as symbols used in a specific field.

[0038] <Fourth Embodiment> Next, a fourth embodiment of this disclosure will be described with reference to the drawings. This embodiment shows an outline of the information processing device, etc., described in the embodiments described above. Note that the drawings may be relevant to any of the embodiments.

[0039] First, the hardware configuration of the information processing device 100 in this disclosure will be described. The information processing device 100 is composed of a general information processing device, and as an example, it is equipped with the following hardware configuration as shown in Figure 6. ·CPU(Central Processing Unit)101(Arithmetic unit) ROM (Read Only Memory) 102 (Storage Device) • RAM (Random Access Memory) 103 (Storage Device) • Program group 104 loaded into RAM 103 • Storage device 105 for storing the program group 104 • Drive device 106 for reading and writing to external storage medium 110 of the information processing device. • Communication interface 107 connecting to a communication network 111 outside the information processing device. • Input / output interface 108 for data input and output. • Bus 109 connecting each component

[0040] Figure 8 shows an example of the hardware configuration of the information processing device 100, and the hardware configuration of the information processing device is not limited to the case described above. For example, the information processing device may consist of only a part of the configuration described above, such as not having a drive device 106. In addition, the information processing device may use a GPU (Graphic Processing Unit), DSP (Digital Signal Processor), MPU (Micro Processing Unit), FPU (Floating point number Processing Unit), PPU (Physics Processing Unit), TPU (Tensor Processing Unit), quantum processor, microcontroller, or a combination thereof instead of the CPU described above.

[0041] The information processing device 100 can be equipped with the first linking unit 121, the candidate extraction unit 122, and the second linking unit 123 shown in Figure 9 by having the CPU 101 acquire the program group 104 and execute it. The program group 104 is stored in the storage device 105 or ROM 102 in advance, for example, and the CPU 101 loads it into the RAM 103 and executes it as needed. The program group 104 may also be supplied to the CPU 101 via the communication network 111, or it may be stored in the storage medium 110 in advance, and the drive device 106 reads the program and supplies it to the CPU 101. However, the first linking unit 121, the candidate extraction unit 122, and the second linking unit 123 described above may be constructed with dedicated electronic circuits for realizing such means.

[0042] The first linking unit 121 associates the second entity with the first entity based on a first knowledge graph consisting of a first entity in a first form notation and a first relation linking the first entities, a second form notation of the first knowledge graph, and a second knowledge graph consisting of a second entity in a second form notation and a second relation linking the second entities (step S101 in Figure 10). The candidate extraction unit 122 extracts candidate second entities from the second knowledge graph that are candidates for the second entity to be associated with the non-linked first entity, based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and a linked first entity, which is another first entity, that is linked by the first relation (step S102 in Figure 10). The second linking unit 123 associates the extracted candidate second entity with the non-linked first entity (step S103 in Figure 10).

[0043] In the above configuration, the information processing device 100 first associates entities based on the first and second formal representations of the first knowledge graph and the second knowledge graph of the second formal representation. Subsequently, the information processing device 100 searches the second knowledge graph starting from the already associated first entities for unlinked entities of the first knowledge graph that could not be associated, and extracts second entities that are candidates for linking. Then, the information processing device 100 associates the entities of the second knowledge graph that are candidates for linking with the unlinked entities of the first knowledge graph. This enables linking between knowledge graphs and makes it easy to increase the amount of information in the knowledge graphs.

[0044] Furthermore, at least one of the functions of the first linking unit 121, the candidate extraction unit 122, and the second linking unit 123 described above may be performed on an information processing device installed and connected at any location on the network, that is, it may be performed using so-called cloud computing.

[0045] Furthermore, the aforementioned programs can be stored and supplied to a computer using various types of non-transitory computer-readable media. Non-transitory computer-readable media include various types of tangible storage media. Examples of non-transitory computer-readable media include magnetic recording media (e.g., flexible disks, magnetic tapes, hard disk drives), magneto-optical recording media (e.g., magneto-optical disks), CD-ROMs (Read Only Memory), CD-Rs, CD-R / Ws, and semiconductor memory (e.g., mask ROMs, PROMs (Programmable ROMs), EPROMs (Erasable PROMs), flash ROMs, and RAMs (Random Access Memory)). Programs may also be supplied to a computer using various types of transient computer-readable media. Examples of transient computer-readable media include electrical signals, optical signals, and electromagnetic waves. Transitory computer-readable media can be supplied to a computer via wired communication channels such as electric wires and optical fibers, or via wireless communication channels.

[0046] Although the present disclosure has been described above with reference to embodiments, the present disclosure is not limited to the embodiments described above. Various modifications to the structure and details of the present disclosure are possible, as can be understood by those skilled in the art within the scope of the present disclosure. Furthermore, each of the embodiments described above can be combined with other embodiments as appropriate.

[0047] <Note> Some or all of the above embodiments may also be described as follows. The general configuration of the information processing apparatus, information processing method, and program in this disclosure is described below. However, this disclosure is not limited to the configurations described below. Furthermore, some or all of the configurations and functions described in Appendices 2 to 8, which are dependent on Appendice 1 below, may also be dependent on Appendices 9 and 10 in the same way as Appendices 2 to 8. Moreover, not limited to Appendices 1, 9, and 10, some or all of the configurations and functions described as appendices may also be dependent on similar hardware, software, various recording means for recording software, or systems, without departing from the embodiments described above. (Note 1) A first linking unit that associates the second entity with the first entity based on a first knowledge graph consisting of first entities in a first form of notation and first relations connecting the first entities, a second form of notation of the first knowledge graph, and a second knowledge graph consisting of second entities in a second form of notation and second relations connecting the second entities, A candidate extraction unit extracts candidate second entities from the second knowledge graph that are candidates for the second entity to be associated with the non-linked first entity, based on a linked first entity, which is another first entity connected by the first relation to a predetermined non-linked first entity, which is a first entity in the first knowledge graph that is not associated with the second entity; A second linking unit that associates the extracted candidate second entity with the first non-linked entity, Equipped with an information processing device. (Note 2) The information processing device described in Appendix 1, The candidate extraction unit extracts the candidate second entity from the second knowledge graph by searching the second knowledge graph starting from the second entity corresponding to the first link entity. Information processing device. (Note 3) The information processing device described in Appendix 1, The second linking unit calculates the degree of relevance between the extracted candidate second entity and the data in the data file related to the first knowledge graph, and associates the candidate second entity with the unlinked first entity based on the degree of relevance. Information processing device. (Note 4) The information processing device described in Appendix 2, The candidate extraction unit extracts the candidate second entity from the second knowledge graph based on the link first entity and the first relation linked to the link first entity. Information processing device. (Note 5) The information processing device described in Appendix 4, The candidate extraction unit searches for and extracts the candidate second entity from the second knowledge graph based on search rules corresponding to the first link entity and the first relation linked to the first link entity. Information processing device. (Note 6) The information processing device described in Appendix 3, The second linking unit calculates the degree of relevance based on the similarity between the extracted candidate second entity and the first entity associated with the second entity in the data file. Information processing device. (Note 7) The information processing device described in Appendix 1, The first linking unit associates the first entity in the first knowledge graph with the second entity in the second knowledge graph whose representation of the first entity matches either the first form or the second form. Information processing device. (Note 8) The information processing device described in Appendix 1, The system includes a conversion unit that extracts the first knowledge graph from text written in the first language, which is the first form of notation, and converts the first knowledge graph into the second language, which is the second form of notation. Information processing device. (Note 9) Information processing device, Based on a first knowledge graph consisting of first entities in first form notation and first relations connecting said first entities, a second form notation of said first knowledge graph, and a second knowledge graph consisting of second entities in second form notation and second relations connecting said second entities, the second entities are associated with the first entities. Based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and another first entity, which is a linked first entity, that is connected by the first relation, a candidate second entity that can be associated with the non-linked first entity is extracted from the second knowledge graph. The extracted candidate second entity is associated with the first non-linked entity. Information processing methods. (Note 10) In an information processing device, Based on a first knowledge graph consisting of first entities in first form notation and first relations connecting said first entities, a second form notation of said first knowledge graph, and a second knowledge graph consisting of second entities in second form notation and second relations connecting said second entities, the second entities are associated with the first entities. Based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and another first entity, which is a linked first entity, that is connected by the first relation, a candidate second entity that can be associated with the non-linked first entity is extracted from the second knowledge graph. The extracted candidate second entity is associated with the first non-linked entity. A program to execute a process. [Explanation of Symbols]

[0048] 10 Information Processing Devices 11 Extraction part 12 Translation Department 13 Entity Linking Section 14. Non-linked entity extraction section 15. Search Origin Entity Extraction Unit 16 Linking candidate extraction unit 17. Relevance Calculation Unit 18 Non-linked entity linking section 21 Search Rule Storage Unit 22 Graph data storage unit 100 Information Processing Devices 101 CPU 102 ROM 103 RAM 104 Program Groups 105 Storage device 106 Drive unit 107 Communication Interface 108 Input / Output Interfaces 109 Bus 110 Storage medium 111 Communication Network 121 First linking section 122 Candidate Extraction Unit 123 Second linking section

Claims

1. A first linking unit that associates the second entity with the first entity based on a first knowledge graph consisting of first entities in a first form of notation and first relations connecting the first entities, a second form of notation of the first knowledge graph, and a second knowledge graph consisting of second entities in a second form of notation and second relations connecting the second entities, A candidate extraction unit extracts candidate second entities from the second knowledge graph that are candidates for the second entity to be associated with the non-linked first entity, based on a linked first entity, which is another first entity connected by the first relation to a predetermined non-linked first entity, which is a first entity in the first knowledge graph that is not associated with the second entity; A second linking unit that associates the extracted candidate second entity with the first non-linked entity, Equipped with an information processing device.

2. An information processing apparatus according to claim 1, The candidate extraction unit extracts the candidate second entity from the second knowledge graph by searching the second knowledge graph starting from the second entity corresponding to the first link entity. Information processing device.

3. An information processing apparatus according to claim 1, The second linking unit calculates the degree of relevance between the extracted candidate second entity and the data in the data file related to the first knowledge graph, and associates the candidate second entity with the unlinked first entity based on the degree of relevance. Information processing device.

4. An information processing apparatus according to claim 2, The candidate extraction unit extracts the candidate second entity from the second knowledge graph based on the link first entity and the first relation linked to the link first entity. Information processing device.

5. An information processing apparatus according to claim 4, The candidate extraction unit searches for and extracts the candidate second entity from the second knowledge graph based on search rules corresponding to the first link entity and the first relation linked to the first link entity. Information processing device.

6. An information processing apparatus according to claim 3, The second linking unit calculates the degree of relevance based on the similarity between the extracted candidate second entity and the first entity associated with the second entity in the data file. Information processing device.

7. An information processing apparatus according to claim 1, The first linking unit associates the first entity in the first knowledge graph with the second entity in the second knowledge graph whose representation of the first entity matches either the first form or the second form. Information processing device.

8. An information processing apparatus according to claim 1, The system includes a conversion unit that extracts the first knowledge graph from text written in the first language, which is the first form of notation, and converts the first knowledge graph into the second language, which is the second form of notation. Information processing device.

9. Information processing device, Based on a first knowledge graph consisting of first entities in first form notation and first relations connecting said first entities, a second form notation of said first knowledge graph, and a second knowledge graph consisting of second entities in second form notation and second relations connecting said second entities, the second entities are associated with the first entities. Based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and another first entity, which is a linked first entity, that is connected by the first relation, a candidate second entity that can be associated with the non-linked first entity is extracted from the second knowledge graph. The extracted candidate second entity is associated with the first non-linked entity. Information processing methods.

10. In an information processing device, Based on a first knowledge graph consisting of first entities in first form notation and first relations connecting said first entities, a second form notation of said first knowledge graph, and a second knowledge graph consisting of second entities in second form notation and second relations connecting said second entities, the second entities are associated with the first entities. Based on a non-linked first entity, which is a predetermined first entity in the first knowledge graph that is not associated with the second entity, and another first entity, which is a linked first entity, that is connected by the first relation, a candidate second entity that can be associated with the non-linked first entity is extracted from the second knowledge graph. The extracted candidate second entity is associated with the first non-linked entity. A program to execute a process.

Citation Information

Patent Citations

  • Healthcare risk estimation system and method

    JP2017174406A