Method and System for Judging Similarity of Military Scientific Research and Production Data Based on Knowledge Graph

Abstract

An embodiment of the present invention provides a method and system for discriminating the similarity of military scientific research and production data based on a knowledge graph, including constructing an ontology system, including constructing an ontology for generating military scientific research and production data and a first association relationship between ontologies, wherein the ontology includes various elements; constructing a knowledge graph, including extracting entities and a second association relationship between the entities included in each original file from the original files based on the ontology system, and generating a knowledge graph corresponding to the original file after entity alignment and association relationship reasoning, wherein the entity is at least one of the various elements of the ontology; constructing a knowledge subgraph, including extracting the quantity information of the entities and the quantity information of the second association relationship in each file to be discriminated based on the knowledge graph, and determining the knowledge subgraph of each file to be discriminated; and discriminating the similarity of the files to be discriminated based on the knowledge subgraph of each file to be discriminated.

Description

Technical Field

[0001] The present invention relates to the field of data processing, and particularly to a method and system for discriminating the similarity of military scientific research and production data based on a knowledge graph. Background Art

[0002] During the long-term military scientific research and production process, a large number of data files have been accumulated. When carrying out related work such as data processing and statistical analysis, there is a problem of duplicate data files, that is, multiple files have different description angles, analysis levels, and production units, but essentially describe the same content. In this case, when carrying out data processing, merging and integration are required, and when carrying out statistical analysis, statistics should be carried out according to one item rather than multiple items.

[0003] Currently, the problem of detecting duplicate similar files mainly adopts the manual method, and experienced senior experts judge whether the essential content of multiple files is the same. This method is time-consuming and laborious, and the amount of data that can be processed is limited, making it difficult to meet the demand for rapid processing of large-scale unstructured data under the wave of digital and intelligent development.

[0004] From a technical perspective, methods such as word repetition rate calculation, TF-IDF keyword calculation, and text vector cosine similarity calculation are generally adopted. The word repetition rate calculation method refers to simply counting the proportion of repeated words between multiple data files to the total number of words. The higher the proportion, the stronger the similarity; the TF-IDF keyword calculation method takes words as basic units, calculates the frequency of keyword occurrences and the frequency of occurrences in all documents, and counts the topic words in the data files. The closer the topic words are, the stronger the similarity; the text vector cosine similarity calculation method refers to constructing text vectors and calculating the cosine similarity between the text vectors. The higher the cosine similarity, the stronger the similarity.

[0005] The prior art only uses text frequency and structure as the basis for similarity judgment, lacking semantic analysis of data files, resulting in difficulty in significantly improving the accuracy of similarity discrimination. Some content with different description methods but similar essential content cannot be recognized at all. Summary of the Invention

[0006] The embodiments of the present invention provide a method and system for discriminating the similarity of military scientific research and production data based on a knowledge graph, which constructs an ontology system based on a domain, and then constructs a knowledge graph based on the ontology system. Furthermore, for the data files whose similarity needs to be discriminated, a knowledge subgraph is constructed from the knowledge graph, and the number of repeated entities and association relationships in the knowledge subgraph is calculated to calculate the similarity of the data files.

[0007] The embodiments of the present invention provide a method for discriminating the similarity of military scientific research and production data based on a knowledge graph, including:

[0008] Construct an ontology system, including constructing an ontology that generates military industrial scientific research and production data and a first association relationship between the ontologies, where the ontology includes multiple elements;

[0009] Construct a knowledge graph, including extracting entities and a second association relationship between the entities included in each original document from the original documents based on the ontology system, and generating a knowledge graph corresponding to the original document after entity alignment and association relationship reasoning, where the entity is at least one of the multiple elements of the ontology;

[0010] Construct a knowledge subgraph, including extracting the quantity information of entities and the quantity information of the second association relationship of each entity in the to-be-discriminated document based on the knowledge graph, and determining the knowledge subgraph of each to-be-discriminated document;

[0011] Discriminate the similarity of the to-be-discriminated documents based on the knowledge subgraph of each to-be-discriminated document.

[0012] In some embodiments of the present invention, the ontology includes at least one of an organization, equipment, personnel, task, time, and location;

[0013] The first association relationship is an association relationship formed by any two elements among multiple elements in the ontology, or an association relationship between different entities corresponding to the same element.

[0014] In some embodiments of the present invention, the method for discriminating the similarity of military industrial scientific research and production data based on the knowledge graph further includes:

[0015] Update the ontology used to construct the ontology system and the first association relationship between the ontologies based on a preset interval duration.

[0016] In some embodiments of the present invention, each of the second association relationships includes two entities.

[0017] In some embodiments of the present invention, if the to-be-discriminated documents are to-be-discriminated document a and to-be-discriminated document b respectively, the formula for discriminating the similarity of the to-be-discriminated documents is as follows:

[0018]

[0019] Among them,

[0020] S ab respectively represents the similarity between the knowledge subgraph a of the to-be-discriminated document a and the knowledge subgraph b of the to-be-discriminated document b;

[0021] e a is the quantity information of entities included in the knowledge subgraph a, and e b is the quantity information of entities included in the knowledge subgraph b;

[0022] e ab is the quantity information of duplicate entities included in knowledge sub - graph a and knowledge sub - graph b;

[0023] r a is the quantity information of the second association relationship included in knowledge sub - graph a, r b is the quantity information of the second association relationship included in knowledge sub - graph b;

[0024] r ab is the quantity information of duplicate second association relationships included in knowledge sub - graph a and knowledge sub - graph b.

[0025] An embodiment of the present invention also provides a similarity discrimination system for military industrial scientific research and production data based on a knowledge graph, including:

[0026] An ontology system construction module, which is used to construct an ontology system, including constructing an ontology for generating military industrial scientific research and production data and a first association relationship between ontologies, wherein the ontology includes multiple elements;

[0027] A knowledge graph construction module, which is used to construct a knowledge graph, including extracting entities and a second association relationship between entities included in each original file from the original files based on the ontology system, and generating a knowledge graph corresponding to the original file after entity alignment and association relationship reasoning, wherein the entity is at least one of the multiple elements of the ontology;

[0028] A knowledge sub - graph construction module, which is used to construct a knowledge sub - graph, including extracting the quantity information of entities and the quantity information of the second association relationship of each entity in the file to be discriminated from the knowledge graph, and determining the knowledge sub - graph of each file to be discriminated;

[0029] A discrimination module, which is used to discriminate the similarity of the file to be discriminated based on the knowledge sub - graph of each file to be discriminated.

[0030] In some embodiments of the present invention, the ontology includes at least one of an institution, equipment, personnel, task, time, and location;

[0031] The first association relationship is an association relationship formed by any two elements among multiple elements in the ontology, or an association relationship between different entities corresponding to the same element.

[0032] In some embodiments of the present invention, the similarity discrimination system for military industrial scientific research and production data based on a knowledge graph further includes:

[0033] An update module, which is used to update the ontology for constructing the ontology system and the first association relationship between ontologies based on a preset interval duration.

[0034] In some embodiments of the present invention, each of the second association relationships includes two entities.

[0035] In some embodiments of the present invention, if the files to be discriminated are file a to be discriminated and file b to be discriminated respectively, the formula for the discrimination module to discriminate the similarity of the files to be discriminated is as follows:

[0036]

[0037] Wherein,

[0038] S ab represents the similarity between knowledge sub-graph a of file a to be discriminated and knowledge sub-graph b of file b to be discriminated respectively;

[0039] e a is the quantity information of entities included in knowledge sub-graph a, and e b is the quantity information of entities included in knowledge sub-graph b;

[0040] e ab is the quantity information of duplicate entities included in knowledge sub-graph a and knowledge sub-graph b;

[0041] r a is the quantity information of the second association relationships included in knowledge sub-graph a, and r b is the quantity information of the second association relationships included in knowledge sub-graph b;

[0042] r ab is the quantity information of duplicate second association relationships included in knowledge sub-graph a and knowledge sub-graph b.

[0043] The method and system for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph provided by the embodiments of the present invention have the following advantages: It constructs an ontology system in the field of military industrial scientific research and production, solidifies the experience of domain experts, clarifies the ontologies and association relationships such as institutions, equipment, personnel, tasks, time, locations, etc., and provides knowledge support for semantic-based similarity discrimination; at the same time, it applies knowledge graph construction technology to extract entities and relationships from data files, constructs knowledge sub-graphs, and maps them into the global knowledge graph, represents the unstructured text from the semantic perspective, and more accurately represents the essential content of different data files; furthermore, it discriminates the repetition ratio of entities and relationships in the sub-graphs constructed by different documents as the basis for similarity discrimination. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 is a flowchart of the method for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph according to an embodiment of the present invention;

[0045] Figure 2Schematic flowchart of the method for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph according to an embodiment of the present invention. Specific implementation manners

[0046] To enable those skilled in the art to better understand the technical solutions of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and specific implementation manners.

[0047] In this specification, phrases such as "in one embodiment", "in another embodiment", "in yet another embodiment", "in an embodiment", "in some embodiments", or "in other embodiments" may all refer to one or more of the same or different embodiments according to the present invention.

[0048] Hereinafter, specific embodiments of the present invention will be described with reference to the accompanying drawings; however, it should be understood that the embodiments of the present invention are merely embodiments of the present invention, and it can be implemented in various ways. Well-known and / or repetitive functions and structures are not described in detail to clarify the true intention based on the user's historical operations, and to avoid unnecessary or redundant details from obscuring the present invention. Therefore, the specific structural and functional details of the present invention are not intended to be limiting, but only as a basis for claims and a representative basis for teaching those skilled in the art to use the present invention diversely in substantially any suitable detailed structure.

[0049] An embodiment of the present invention provides a method for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph, including:

[0050] Step 1: Construct an ontology system, including constructing an ontology for generating military industrial scientific research and production data and a first association relationship between the ontologies, where the ontology includes multiple elements. Specifically, it can be based on the existing discrimination of the similarity of scientific research and production data to construct an ontology for military industrial scientific research and production data and a first association relationship between the ontologies. Of course, when constructing an ontology for military industrial scientific research and production data and a first association relationship between the ontologies, the experience of experts in the field of military industrial scientific research and production can also be combined to more fully provide knowledge support in the professional field for the construction of the knowledge graph.

[0051] In this embodiment, the ontology includes at least one of an organization, equipment, personnel, task, time, and location. Correspondingly, the first association relationship may include, for example, organization a produced equipment a, personnel a produced equipment a, personnel a executed task a, the production time of equipment a is time a, equipment a was produced at location a, organization a is an affiliated organization of organization b, etc., which are association relationships formed by any two elements among the organization, equipment, personnel, task, time, and location included in the ontology, or association relationships between different entities corresponding to the same element. Among them, the organization can be a military industrial scientific research unit or other production sites, etc.

[0052] In some embodiments of the present invention, the method for discriminating the similarity of military scientific research and production data based on a knowledge graph further includes: updating the ontology used to construct the ontology system and the first association relationships between the ontologies based on a preset interval duration, so as to be able to update the ontology system in combination with the changes in the ontologies in the ontology system and the first association relationships between the ontologies, to ensure the accuracy of subsequent knowledge graph construction.

[0053] Step 2: Construct a knowledge graph, including extracting entities and the second association relationships between the entities included in each original file from the original files based on the ontology system, and generating the knowledge graph corresponding to the original files after entity alignment and association relationship reasoning, where the entity is at least one of multiple elements of the ontology, the original file may be a data file, and the data file may be a file in text form or a file in picture form. If it is a file in picture form, the file in picture form also needs to be recognized, and similarity discrimination is performed based on the recognized text information.

[0054] Step 3: Construct a knowledge subgraph, including extracting the quantity information of the entities and the quantity information of the second association relationships of each entity in the file to be discriminated from the knowledge graph, and determining the knowledge subgraph of each file to be discriminated.

[0055] In this embodiment, each of the second association relationships includes two entities. As an example, if there are 2 files to be discriminated, namely file to be discriminated a and file to be discriminated b, the entities corresponding to file to be discriminated a are organization a and equipment a produced by organization a, while the entities corresponding to file to be discriminated b are organization a, equipment a produced by organization a, organization b, and in addition, there is person a who produces equipment a.

[0056] Furthermore, in the above example, the quantity information of the entities in file to be discriminated a is 2, and the quantity information of the second association relationship is 1. In file to be discriminated b, the quantity information of the entities is 4, that is, organization a, equipment a, organization b, person a, and the quantity information of the second association relationship is 3, that is, respectively, equipment a produced by organization a, equipment a produced by person a, and person a of organization a. At the same time, the quantity information of the same entities in file to be discriminated a and file to be discriminated b is 2, and the quantity information of the repeated second association relationships is 1.

[0057] Step 4: Discriminate the similarity of the files to be discriminated based on the knowledge subgraph of each original file.

[0058] In some embodiments of the present invention, if the files to be discriminated are file to be discriminated a and file to be discriminated b respectively, the formula for discriminating the similarity of the files to be discriminated is as follows:

[0059]

[0060] Among them,

[0061] S ab respectively represents the similarity between knowledge sub-graph a of the file a to be discriminated and knowledge sub-graph b of the file b to be discriminated;

[0062] e a is the quantity information of entities included in knowledge sub-graph a, and e b is the quantity information of entities included in knowledge sub-graph b;

[0063] e ab is the quantity information of duplicate entities included in knowledge sub-graph a and knowledge sub-graph b;

[0064] r a is the quantity information of the second association relationships included in knowledge sub-graph a, and r b is the quantity information of the second association relationships included in knowledge sub-graph b;

[0065] r ab is the quantity information of duplicate second association relationships included in knowledge sub-graph a and knowledge sub-graph b.

[0066] Combining the above embodiments, that is, taking the quantity information of entities of the file a to be discriminated as 2, the quantity information of the second association relationships as 1, the quantity information of entities of the file b to be discriminated as 4, the quantity information of the second association relationships as 3, and at the same time, the quantity information of the same entities in the file a to be discriminated and the file b to be discriminated as 2, and the quantity information of duplicate second association relationships as 1 as an example, then:

[0067]

[0068] It can be seen from the above technical solutions that it constructs an ontology system in the field of military scientific research and production, solidifies the experience of domain experts, clarifies the ontologies and association relationships such as institutions, equipment, personnel, tasks, time, location, etc., and provides knowledge support for semantic similarity discrimination; at the same time, applying the knowledge graph construction technology, entities and relationships are extracted from data files, knowledge sub-graphs are constructed, and mapped into the global knowledge graph, and knowledge representation of unstructured text is carried out from the semantic perspective, more accurately representing the essential content of different data files; furthermore, the repetition ratio of entities and relationships of sub-graphs constructed from different documents is discriminated as the basis for similarity discrimination.

[0069] It can well solve the problems existing in the data management and statistical analysis work in the field of existing military scientific research and production, that is, it is necessary to process a large amount of unstructured data. The traditional manual processing method is time-consuming and laborious and the processing quantity is limited. At the same time, for data files with different expression forms but the same essential content, similarity discrimination methods need to be used to mark them as duplicates and should not be counted multiple times. However, the existing text frequency-based methods are difficult to analyze from the semantic level, and the accuracy of using computers for similarity discrimination is not high, resulting in great difficulty in application implementation.

[0070] This paper adopts a similarity discrimination method for military scientific research and production data based on knowledge graph. First, domain experts construct an ontology system to solidify expert experience. Second, a domain knowledge graph is constructed for the training set using the knowledge graph construction method. Third, for the data files whose similarity needs to be discriminated, knowledge subgraphs are quickly constructed in combination with the knowledge graph. Finally, the number of repeated entities and relationships in the knowledge subgraphs is calculated to calculate the similarity of the data files.

[0071] The embodiment of the present invention also provides a similarity discrimination system for military scientific research and production data based on knowledge graph, including:

[0072] An ontology system construction module, which is used to construct an ontology system, including constructing the ontology that generates military scientific research and production data and the first association relationship between ontologies. Among them, the ontology includes multiple elements;

[0073] A knowledge graph construction module, which is used to construct a knowledge graph, including extracting entities from the original files and the second association relationship between the entities included in each original file based on the ontology system, and generating the knowledge graph corresponding to the original file after entity alignment and association relationship reasoning. Among them, the entity is at least one of the multiple elements of the ontology;

[0074] A knowledge subgraph construction module, which is used to construct a knowledge subgraph, including extracting the quantity information of entities and the quantity information of the second association relationship of each entity in the file to be discriminated based on the knowledge graph, and determining the knowledge subgraph of each file to be discriminated;

[0075] A discrimination module, which is used to discriminate the similarity of the files to be discriminated based on the knowledge subgraphs of each file to be discriminated.

[0076] In some embodiments of the present invention, the ontology includes at least one of organization, equipment, personnel, task, time and location;

[0077] The first association relationship is the association relationship formed by any two elements among multiple elements in the ontology, or the association relationship between different entities corresponding to the same element.

[0078] In some embodiments of the present invention, the similarity discrimination system for military scientific research and production data based on a knowledge graph further includes:

[0079] An update module, which is used to update the ontology for constructing the ontology system and the first association relationships between ontologies based on a preset interval duration.

[0080] In some embodiments of the present invention, each of the second association relationships includes two entities.

[0081] In some embodiments of the present invention, if the files to be discriminated are file a to be discriminated and file b to be discriminated respectively, the formula for the discrimination module to discriminate the similarity of the files to be discriminated is as follows:

[0082]

[0083] Where

[0084] S ab respectively represents the similarity between knowledge sub-graph a of the file a to be discriminated and knowledge sub-graph b of the file b to be discriminated;

[0085] e a is the quantity information of entities included in knowledge sub-graph a, and e b is the quantity information of entities included in knowledge sub-graph b;

[0086] e ab is the quantity information of repeated entities included in knowledge sub-graph a and knowledge sub-graph b;

[0087] r a is the quantity information of the second association relationships included in knowledge sub-graph a, and r b is the quantity information of the second association relationships included in knowledge sub-graph b;

[0088] r ab is the quantity information of repeated second association relationships included in knowledge sub-graph a and knowledge sub-graph b.

[0089] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be recognized that the above description should not be considered as a limitation of the present invention. After those skilled in the art have read the above content, various modifications and substitutions to the present invention will be obvious. Therefore, the protection scope of the present invention should be defined by the appended claims.

Claims

1. A method for discriminating the similarity of military scientific research and production data based on a knowledge graph, characterized in that, Including: Construct an ontology system, including constructing an ontology that generates military industrial scientific research and production data and a first association relationship between the ontologies, where the ontology includes multiple elements; Construct a knowledge graph, including extracting entities and a second association relationship between the entities included in each original document from the original documents based on the ontology system, and generating a knowledge graph corresponding to the original document after entity alignment and association relationship reasoning, where the entity is at least one of the multiple elements of the ontology; Construct a knowledge sub-graph, including extracting the quantity information of the entities and the quantity information of the second association relationship in each to-be-discriminated document from the knowledge graph, and determining the knowledge sub-graph of each to-be-discriminated document; Based on the knowledge sub-graph of each to-be-discriminated document, discriminate the similarity of the to-be-discriminated document; If the to-be-discriminated documents are to-be-discriminated document a and to-be-discriminated document b respectively, the formula for discriminating the similarity of the to-be-discriminated documents is as follows: Wherein, S ab respectively represent the similarity between the knowledge sub-graph a of the file a to be discriminated and the knowledge sub-graph b of the file b to be discriminated; e a is the quantity information of entities included in knowledge sub-graph a, e b is the quantity information of entities included in knowledge sub-graph b; e ab It is the quantity information of the repeated entities included in knowledge subgraph a and knowledge subgraph b; r a is the quantity information of the second association relationship included in knowledge subgraph a, r b is the quantity information of the second association relationship included in knowledge subgraph b; r ab It is the quantity information that the knowledge subgraph a and the knowledge subgraph b contain duplicate second association relationships.

2. The method for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph according to claim 1, characterized in that The ontology includes at least one of organization, equipment, personnel, task, time, and location; The first association relationship is an association relationship formed by any two elements among multiple elements in the ontology, or an association relationship between different entities corresponding to the same element.

3. The method for discriminating the similarity of military industrial scientific research and production data based on the knowledge graph according to claim 2, wherein, It further includes: Based on a preset interval duration, update the ontology used to construct the ontology system and the first association relationship between the ontologies.

4. The method for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph according to claim 3, characterized in that Each of the second association relationships includes two entities.

5. A military scientific research and production data similarity discrimination system based on a knowledge graph, characterized in that, Including: An ontology system construction module, which is used to construct an ontology system, including constructing an ontology that generates military industrial scientific research and production data and a first association relationship between the ontologies, where the ontology includes multiple elements; A knowledge graph construction module, which is used to construct a knowledge graph, including extracting entities and a second association relationship between the entities included in each original document from the original documents based on the ontology system, and generating a knowledge graph corresponding to the original document after entity alignment and association relationship reasoning, where the entity is at least one of the multiple elements of the ontology; A knowledge sub-graph construction module, which is used to construct a knowledge sub-graph, including extracting the quantity information of the entities and the quantity information of the second association relationship in each to-be-discriminated document from the knowledge graph, and determining the knowledge sub-graph of each to-be-discriminated document; A discrimination module, which is used to discriminate the similarity of the to-be-discriminated document based on the knowledge sub-graph of each to-be-discriminated document; If the to-be-discriminated documents are to-be-discriminated document a and to-be-discriminated document b respectively, the formula for the discrimination module to discriminate the similarity of the to-be-discriminated documents is as follows: Wherein, S ab respectively represent the similarity between the knowledge subgraph a of the file a to be discriminated and the knowledge subgraph b of the file b to be discriminated; e a is the quantity information of entities included in knowledge sub-graph a, e b is the quantity information of entities included in knowledge sub-graph b; e ab is the quantity information of the repeated entities included in knowledge sub-graph a and knowledge sub-graph b; r a is the quantity information of the second association relationship included in knowledge sub-graph a, r b is the quantity information of the second association relationship included in knowledge sub-graph b; r ab It is the quantity information that the knowledge subgraph a and the knowledge subgraph b contain duplicate second association relationships.

6. The system for discriminating the similarity of military industrial scientific research and production data based on a knowledge graph according to claim 5, characterized in that The ontology includes at least one of organization, equipment, personnel, task, time, and location; The first association relationship is an association relationship formed by any two elements among multiple elements in the ontology, or an association relationship between different entities corresponding to the same element.

7. The similarity discrimination system for military industrial scientific research and production data based on a knowledge graph according to claim 6, wherein It further includes: An update module, which is used to update the ontology for constructing the ontology system and the first association relationship between ontologies based on a preset interval duration.

8. The similarity discrimination system for military scientific research and production data based on a knowledge graph according to claim 7, characterized in that Each of the second association relationships includes two entities.

Images