An ontology organization method based on a knowledge graph

Abstract

The application discloses a kind of based on ontology organization method of knowledge graph, belong to knowledge graph technical field.First, element ontology structure and ontology element knowledge graph structure are defined;Through big language model and rdflib library, the entity data information of RDF graph model is extracted and is converted into triple, is stored and forms the structured data layer of knowledge graph, selects core element entity, the name, definition structure feature and graph structure feature are vectorized and the similarity of core entity pair is calculated, traverses and merges entity pair, based on entity pair, the most authoritative value of core element entity node is calculated.The application realizes ontology fine-grained organization and cross ontology accurate alignment by analyzing ontology semantic element, effectively improves the automation degree and accuracy of ontology organization process.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of knowledge graph, and particularly relates to an ontology organization method based on a knowledge graph. BACKGROUND

[0002] With the remarkable application effect of artificial intelligence technology in small field knowledge services, large and complex fields such as agriculture, medicine and construction have also begun to explore intelligent knowledge services. However, due to the characteristics of wide interdisciplinary, many subdivided fields and complex knowledge range of these fields, higher requirements are put forward for the organization ability of massive knowledge.

[0003] Ontology is a formal and normative representation of the conceptualization of knowledge in a specific field. Its core is to realize the structured organization of field knowledge by constructing a clear concept framework, semantic relationship between concepts and related attributes. As an important tool for knowledge organization, ontology can promote cross-disciplinary and cross-field knowledge sharing and reuse by constructing a clear concept framework and semantic relationship.

[0004] Existing scholars and related organizations have carried out ontology evaluation and ontology community development work:

[0005] In the field of biomedicine, Hoehndorf et al. evaluated five ontologies including Gene Ontology, Human Disease Ontology, Foundational Model of Anatomy, NDF-RT ontology and ICD ontology; in the field of agriculture, the performance of multiple agricultural ontologies such as RiceDO and TreatO, pest control ontology and tea tree disease ontology has been evaluated.

[0006] Alexa T. McCray et al. assigned a unique identifier CUI to each concept and associated different concepts using multiple relationship types, integrating 60 vocabulary tables in the medical field; AgroPortal developed by INRA provides ontology resource information through methods such as category classification, group division and semantic mapping, and designs detailed metadata information to support automatic retrieval and semantic query; Smith B et al. constructed the open biology community “OBO Foundry”, which organizes ontologies through methods such as integration and classification, registry and association link, semantic link and cross-ontology relationship, and develops visualization tools and dashboards to facilitate researchers to intuitively refer to ontologies.

[0007] However, the above existing methods still have the following shortcomings: (1) the existing ontology organization methods are mostly limited to the classification, retrieval and simple integration of ontologies, lacking systematic analysis and organization of the internal semantic elements of the ontology, making it difficult to achieve fine-grained semantic organization and deep understanding of ontology resources, and not fully utilizing the maximum value of existing ontology resources; (2) the existing ontology organization methods fail to realize automatic analysis and extraction of ontology resources, and the organization process relies on manual intervention, which is low in efficiency and easily affected by human factors, and is difficult to adapt to the rapid growth of ontology scale and quantity; (3) the existing ontology organization methods generally lack unified representation specifications and semantic alignment methods between ontologies, which cannot effectively support knowledge integration, association and reasoning between ontologies, resulting in that the value of existing ontology resources is not fully tapped and utilized. SUMMARY

[0008] In order to solve the above problems in the prior art, the present application realizes the following technical scheme:

[0009] The present application provides an ontology organization method based on a knowledge graph, comprising the following steps:

[0010] Step 1: defining the element ontology structure and the ontology element knowledge graph structure;

[0011] Step 2: knowledge extraction of literature-based ontology, extracting entity data information of the literature-based ontology conforming to the element ontology structure and converting it into a triple form; knowledge extraction of OWL-based ontology, extracting entity data information of the OWL-based ontology conforming to the element ontology structure and converting it into a triple form;

[0012] Step 3: storing the entity data information of the literature-based ontology in triple form and the entity data information of the OWL-based ontology in triple form into the ontology element knowledge graph, forming a structured data layer of the ontology element knowledge graph;

[0013] Step 4: selecting core element entities of the two types of ontologies in the structured data layer of the ontology element knowledge graph, vectorizing the multiple features of the core element entities respectively to obtain the vectorized core element entity features, and calculating the similarity of the vectorized core element entity features of the literature-based ontology and the vectorized core element entity features of the OWL-based ontology to obtain core element entity pairs;

[0014] Step 5: traversing the core element entity pairs, merging the core element entity pairs with similar mapping relationships, obtaining core element entity nodes and recording the number of times the core element entity nodes are mapped, adding the number of times the core element entity nodes are mapped to the authority value at the time of storage to obtain the final authority value of the core element entity nodes.

[0015] In step 2, knowledge extraction is performed on the literature ontology, and entity data information of the literature ontology conforming to the element ontology structure is extracted and converted into triple form, including:

[0016] The task target of literature ontology knowledge extraction, the standard of knowledge extraction, and the example of structured triple are integrated into a Prompt string. The Prompt string is combined with the literature ontology as input, and the large language model of ChatGPT-4o is called by using API as an interface. Knowledge extraction is performed on the literature ontology, and entity data information of the literature ontology conforming to the element ontology structure is extracted and converted into triple form and output.

[0017] In step 2, knowledge extraction is performed on the OWL ontology, and entity data information of the OWL ontology conforming to the element ontology structure is extracted and converted into triple form, including:

[0018] The RDF graph model is constructed by loading the OWL ontology through the rdflib library.

[0019] According to the label in the RDF graph model, the entity data information of the OWL ontology conforming to the element ontology structure is extracted and converted into triple form.

[0020] In step 3, the carrier of the structured data layer of the ontology element knowledge graph is the Neo4j graph database.

[0021] In step 4, the core element entities of the two types of ontologies are Class entities, ObjectProperty entities, and DataProperty entities.

[0022] In step 4, the multiple features of the core element entities are the name features representing the names of the core element entities, the definition structure features representing the association structures formed by the core element entities through triple form, and the graph structure features representing the structured relationships of the core element entities in the knowledge graph.

[0023] The all-MiniLM-L6-v2 model in the Sentence Transformer framework is used to vectorize the name features and the definition structure features, and the TransE model is used to vectorize the graph structure features.

[0024] In step 4, the core element entity features of the literature ontology after vectorization and the core element entity features of the OWL ontology after vectorization are calculated for similarity, and the core element entity pairs are obtained, including:

[0025] The similarity of the core element entity features of the literature ontology after vectorization and the core element entity features of the OWL ontology after vectorization is calculated, and candidate entity pairs greater than the similarity threshold are extracted.

[0026] Before step 4, the core element entities that do not need to be traversed are removed by using a Cypher statement before the similarity of the core element entity features of the literature ontology after vectorization and the core element entity features of the OWL ontology after vectorization is calculated.

[0027] In step 5, the number of times of mapping is added to the authority value when storage is completed to obtain the final authority value of the core element entity node, and the method further comprises the following steps:

[0028] In step 3, the entity data information is stored in the ontology element knowledge graph, all the entity data information in the form of triples is traversed, the same core element entities are merged, and the structured data layer of the ontology element knowledge graph is formed until the storage is completed. The authority value of the core element entity is increased by one each time the core element entity is merged until the authority value when the storage is completed is obtained.

[0029] Compared with the prior art, the beneficial effects of the present application are as follows:

[0030] (1) The element ontology structure model oriented to the ontology core is constructed, the semantic units of the ontology are divided into five types of standard concepts, and the corresponding relationship type and the attribute type are defined, thereby providing standard support for the standardized representation and structured management of the ontology.

[0031] (2) The construction idea and method of the ontology element knowledge graph are proposed, the entity data information and various semantic elements in the ontology are extracted and modeled, and are stored in the form of a graph database.

[0032] (3) The double-strategy automatic extraction method is proposed in view of the difference between the literature ontology resources and the OWL ontology, the natural language processing and the structured analysis technology are combined to realize the efficient extraction and classification of the ontology, and the automation degree in the organization process of the ontology is improved.

[0033] (4) The matching is performed based on the core element entity to support the semantic fusion and relationship establishment between different ontologies, and the authority value calculation mechanism is introduced to identify the core elements with high credibility and representativeness, thereby further improving the accuracy and effectiveness of the cross-ontology alignment.

[0034] The present application will be further described in detail below with reference to the accompanying drawings and embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0035] Figure 1 is a knowledge graph-based ontology organization method flowchart provided by an embodiment of the present application;

[0036] Figure 2 is an rdflib-based OWL class ontology analysis framework diagram provided by an embodiment of the present application. DETAILED DESCRIPTION

[0037] In order to further illustrate the technical means and effects taken by the present application to achieve the predetermined object, the scheme according to the present application is described in detail below in combination with the drawings and specific embodiments.

[0038] The foregoing and other technical contents, features and effects of the present application can be clearly presented in the specific embodiment description below in combination with the drawings. Through the description of the specific embodiments, the technical means and effects taken by the present application to achieve the predetermined object can be understood more deeply and specifically. However, the accompanying drawings are provided for reference and illustration only, and are not intended to limit the technical scheme of the present application.

[0039] It should be noted that, in this document, relational terms such as first and second and the like can only be used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply that there is any such actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a product or device including a list of elements does not exclude other elements not explicitly listed. Without more limitations, the element defined by the statement "including a" does not exclude the presence of additional identical elements in the product or device including the element.

[0040] As shown in Figure 1 is a knowledge graph-based ontology organization method provided by an embodiment of the present application, which comprises the following steps:

[0041] Step 1: Define the element ontology structure and the ontology element knowledge graph structure.

[0042] Ontology is a formal specification of domain conceptualization. In current research, the structure of ontology is often represented by graphs, mathematical structures or logic, etc. In order to meet the structured expression needs of ontology in the process of organization and reuse, an element ontology structure (EO) is designed to provide a structured framework for the organization and reuse of ontology elements.

[0043] The element ontology structure is defined as follows:

[0044] EO = (C, Ac, R, Ar)

[0045] Wherein, EO is an element ontology structure, C is a set of entity concepts in the ontology; R represents a set of relations in the element ontology, Ac and Ar represent the attributes of entity concepts and the attributes of relations in the ontology respectively, that is, Ac and Ar jointly constitute the attribute set in the ontology; wherein, the application defines five concept sets for EO, the concept set is: C={C1, C2, C3, C4, C5}, wherein C1 is a core element entity class concept of the ontology, which refers to the concept of the core element entity constituting the ontology structure, including concepts, relations and attributes; C2 is an ontology information class concept, which refers to an information class concept for describing the attributes of the ontology itself, including ontology name and ontology language; C3 is an ontology source class concept, which refers to an information class concept for describing the source or origin of the ontology, including literature and open source address; C4 is an ontology tool class concept, which refers to a tool and method for supporting the ontology construction process, including ontology construction tools; C5 is an ontology evaluation class concept, which refers to a method and process for measuring and verifying the quality, performance and usability of the ontology, including ontology evaluation methods; R={R1, R2,..., Rn} defines the relations within the C1 class concept and the relations between the C1 class concept and other class concepts; Ac includes the definition, data type and described attributes of the concept, and Ar includes the definition domain and value domain attributes of the relation.

[0046] The ontology element knowledge graph (OEKG) is used for formal modeling and semantic representation of the core elements and related knowledge information of the ontology resource, and effectively organizes and manages the ontology. The data layer of the ontology element knowledge graph is in the form of triples, and stores specific data information.

[0047] The ontology element knowledge graph structure is defined as follows:

[0048] OEKG = {E, RS, T}

[0049] Wherein, OEKG is a knowledge graph, E represents a set of entity names {e1, e2... eE}, e is the name of an entity, RS represents a set of relations {r1, r2,..., rR}, the relation r is an edge in the knowledge graph, representing the connection between different entities, and T represents a set of triples in the form of entities {t1, t2,..., tT}, each entity is defined as a triple (h, r, t) ∈ T, wherein h represents the head entity, r represents the relation, and t represents the tail entity.

[0050] Step 2: Knowledge extraction is performed on the literature ontology, entity data information of the literature ontology conforming to the element ontology structure is extracted and converted into a triple form; knowledge extraction is performed on the OWL ontology, entity data information of the OWL ontology conforming to the element ontology structure is extracted and converted into a triple form.

[0051] The knowledge extraction of the literature ontology is performed to extract entity data information of the literature ontology conforming to the element ontology structure and convert the entity data information into a triple form, including:

[0052] The task target of the knowledge extraction of the literature ontology, the standard of the knowledge extraction, and the example of the structured triple are integrated into a Prompt string, the Prompt string is combined with the literature ontology as input, the large language model of ChatGPT-4o is called, the API is used as an interface, the knowledge extraction of the literature ontology is performed, entity data information of the literature ontology conforming to the element ontology structure is extracted, and the entity data information is converted into a triple form and output.

[0053] The task target of the knowledge extraction is the specific entity data information identified and extracted in the literature ontology, including two categories. The first category is basic information entities obtained according to the set C2, C3, C4, and C5 of the entity concept in the ontology, including: ontology information entities, ontology source entities, ontology tool entities, and ontology evaluation entities and the association relationship between the entities. The second category is core element entities, including Class entities, ObjectProperty entities, and DataProperty entities and the association relationship between the entities obtained according to the concepts, relationships, and attributes in the concept C1 of the ontology core element entity category.

[0054] The standard of the knowledge extraction conforms to the range of the element ontology structure (EO) on one hand and conforms to the triple form on the other hand.

[0055] The example of the structured triple is used to visually display the specific format of the extraction result, and helps the large language model of ChatGPT-4o to understand the task target of the knowledge extraction and the standard of the knowledge extraction through few-shot learning.

[0056] The knowledge extraction of the OWL ontology is performed to extract entity data information of the OWL ontology conforming to the element ontology structure and convert the entity data information into a triple form, including:

[0057] The RDF graph model is constructed by loading the OWL ontology through the rdflib library;

[0058] The entity data information of the OWL ontology conforming to the element ontology structure is extracted and converted into a triple form according to the labels in the RDF graph model. Based on the characteristics of the OWL ontology itself, the core element entities conforming to the element ontology structure (EO) can be extracted, that is, the Class entities, the ObjectProperty entities, the DataProperty entities, and the association relationship between the entities.

[0059] Specifically, as shown in Figure 2 the rdflib-based OWL ontology analysis framework provided by the embodiment of the present application is shown.

[0060] Traverse all the OWL. Class labels in the RDF graph model, identify and extract the rdfs: label label under each Class entity respectively to obtain the basic information of the corresponding Class entity; for the case of lacking rdfs: label label, a three-level fallback parsing mechanism is adopted, that is, lacking rdfs: label label identifies and extracts skos: prefLabel label, lacking skos: prefLabel label identifies and extracts URI fragment, ensures to obtain the complete basic identification information of Class entity, adopts a dynamic hierarchical calculation algorithm of loop detection, accumulates the inheritance path of the concept class layer by layer, automatically deduces the level of the Class entity, and finally obtains the triple form of the Class entity.

[0061] Traverse all the owl. ObjectProperty labels in the RDF graph model, identify and extract the rdfs: label label, rdfs: subPropertyOf label, rdfs: domain label and the Class entity pointed by the rdfs: range label, build the structured expression between the ObjectProperty entity and the Class entity as the head entity and the tail entity, and obtain the triple form of the ObjectProperty entity.

[0062] Traverse all the owl. DatatypeProperty labels in the RDF graph model, identify and extract the information of rdfs: label label, rdfs: domain label and rdfs: range label, wherein the rdfs: range label corresponds to the data type of the DataProperty entity, and the rdfs: domain label corresponds to the associated Class entity, to obtain the triple form of the DataProperty entity.

[0063] Step 3: Store the entity data information of the literature class ontology in the form of triples and the entity data information of the OWL class ontology in the form of triples into the ontology element knowledge graph, form the structured data layer of the ontology element knowledge graph, and the carrier of the structured data layer of the ontology element knowledge graph is a Neo4j graph database.

[0064] The entity data information of the literature class ontology in the form of triples includes: basic information entities, that is, ontology information class, ontology source class, ontology tool class and ontology evaluation class entities and the association relationship between entities and the association relationship between core element entities; the entity data information of the OWL class ontology in the form of triples includes core element entities and the association relationship between entities.

[0065] In the process of storing entity data information into the ontology element knowledge graph, all entity data information in the form of triples is traversed, and core element entities with the same triple form are merged until the storage is completed, forming a structured data layer of the ontology element knowledge graph. The authority value of the core element entity is increased by one each time the core element entity is merged, until the authority value at the time of storage completion is obtained.

[0066] Step 4: Selecting core element entities of two types of ontologies in the structured data layer of the ontology element knowledge graph, vectorizing the multiple features of the core element entities to obtain vectorized core element entity features, and calculating the similarity of the vectorized core element entity features of the literature ontology and the vectorized core element entity features of the OWL ontology to obtain a pair of core element entities.

[0067] The structured data layer of the ontology element knowledge graph stores entity data information of the literature ontology and the OWL ontology, and the entity data information includes core element entities. In this application, the multiple features of any core element entity are name features representing the name of the core element entity, definition structure features representing the association structure formed by the core element entity through triples, and graph structure features representing the structured relationship of the core element entity in the knowledge graph. Specifically, the name features come from the head entity h or the tail entity t in the triples, the definition structure features come from the relationship r in the triples, and the graph structure features are the entity and relationship network formed by multiple triples.

[0068] The all-MiniLM-L6-v2 model in the Sentence Transformer framework is used to vectorize the name features and the definition structure features. This model is fine-tuned through knowledge distillation technology based on MiniLM, effectively balancing the representation performance and reasoning efficiency, and achieving high-quality vectorization representation. The TransE model is used to vectorize the graph structure features. The TransE model can map the entity and relationship network of the knowledge graph to a low-dimensional vector space, so that the vector representation can accurately reflect the structured semantic association between entities.

[0069] Before calculating the similarity of the vectorized core element entity features of the literature ontology and the vectorized core element entity features of the OWL ontology, Cypher statements are used to eliminate core element entities that do not need to be traversed, in order to improve the traversal efficiency of the core element entities.

[0070] The core element entity features after vectorization of the same type of literature class ontology and the core element entity features after vectorization of the OWL class ontology are subjected to similarity calculation, and candidate entity pairs greater than a similarity threshold value are extracted; wherein the similarity threshold values of the name feature, the definition structure feature and the atlas structure feature are 0.9, 0.8 and 0.7 respectively; the semantic effectiveness of the candidate entity pairs is confirmed, and the candidate entity pairs without actual semantic information, multiple semantic information and semantic irrelevance are removed, to obtain core element entity pairs.

[0071] Step 5: The core element entity pairs are traversed, the core element entity pairs with similar mapping relationships are merged, core element entity nodes are obtained and the mapping times of the core element entity nodes are recorded, the mapping times are added to the authority value at the time of storage, and the final authority value of the core element entity nodes is obtained.

[0072] The authority of the Class entity CE_authority is the number of ontologies with the concept; the authority of the ObjectProperty entity RE_authority refers to the number of ontologies corresponding to the relationships associated with the same head and tail entities; and the authority of the DataProperty entity PE_authority refers to the number of ontologies corresponding to the properties associated with the same head concept entity.

[0073] The core element entity pairs are traversed, the core element entity pairs with similar mapping relationships are merged, core element entity nodes are obtained and the mapping times of the core element entity nodes are recorded, the mapping times are added to the authority value at the time of storage, and the final authority value of the core element entity nodes is obtained.

[0074] The similar mapping relationship includes direct similar mapping and indirect similar mapping, the direct similar mapping is the mapping between the core element entity pairs, and the indirect similar mapping is the association mapping formed through the transitivity of the direct similar mapping.

[0075] The application discloses an ontology organization method based on a knowledge graph, which first defines an element ontology structure and an ontology element knowledge graph structure, then extracts entity data conforming to the structure and converts the entity data into triples by constructing an RDF graph model through a large language model and an rdflib library for a literature class ontology and an OWL class ontology respectively, stores the two types of triples into Neo4j to form a structured data layer, subsequently selects core element entities, i.e. Class entities, ObjectProperty entities and DataProperty entities, respectively vectorizes the name features, the definition structure features and the atlas structure features of the core element entities, calculates the similarities to obtain effective core element entities, merges the core element entity pairs, and obtains the final authority value of the core element entity nodes by adding the mapping times to the authority value at the time of storage.

[0076] The application can accurately identify entity pairs with close semantic association in different ontologies, avoid the limitation of single feature matching, provide multi-dimensional basis for the establishment of cross-ontology semantic association, and reflect the consensus degree of entities in the initial data layer and the association strength in cross-ontology mapping through effective identification of core elements with high credibility and representativeness in the semantic network, thereby further improving the accuracy and effectiveness of cross-ontology alignment. From structure specification, data extraction, association matching to credibility verification, the application comprehensively supports the fine-grained decomposition, structured integration and accurate alignment of cross-ontology semantics of ontology resources.

[0077] The above is a further detailed description of the application in combination with specific preferred embodiments, and the specific implementation of the application cannot be limited to these descriptions. For ordinary skilled persons in the technical field to which the application belongs, without departing from the concept of the application, a number of simple deductions or substitutions can be made, which should be considered as falling within the protection scope of the application.

Claims

1. A method for organizing an ontology based on a knowledge graph, characterized in that, The method comprises the following steps: Step 1: defining element ontology structure and ontology element knowledge graph structure; The element ontology structure comprises a set of entity concepts in the ontology, attributes of the entity concepts in the ontology, a set of relations in the element ontology and attributes of the relations in the ontology; The set of entity concepts in the ontology comprises five categories, namely, ontology core element entity class concepts, ontology information class concepts, ontology source class concepts, ontology tool class concepts and ontology evaluation class concepts; the ontology core element entity class concepts refer to the concepts of core element entities constituting the ontology structure, including concepts, relations and attributes; the ontology information class concepts refer to information class concepts for describing attributes of the ontology itself, including ontology name and ontology language; the ontology source class concepts refer to information class concepts for describing sources or origins of the ontology, including literature and open source addresses; the ontology tool class concepts refer to tools and methods for supporting the ontology construction process, including ontology construction tools; and the ontology evaluation class concepts refer to methods and processes for measuring and verifying the quality, performance and usability of the ontology, including ontology evaluation methods; Step 2: knowledge extraction of literature class ontology, extraction of entity data information of the literature class ontology conforming to the element ontology structure and conversion into a triple form; knowledge extraction of OWL class ontology, extraction of entity data information of the OWL class ontology conforming to the element ontology structure and conversion into a triple form; the entity data information comprises basic information entities and core element entities; the basic information entities comprise ontology information class entities, ontology source class entities, ontology tool class entities and ontology evaluation class entities and associated relations between the entities, which are obtained according to the ontology information class concepts, the ontology source class concepts, the ontology tool class concepts and the ontology evaluation class concepts in the set of entity concepts in the ontology; and the core element entities comprise Class entities, ObjectProperty entities and DataProperty entities and associated relations between the entities, which are obtained according to the concepts, relations and attributes in the ontology core element entity class concepts; Loading the RDF graph model of the OWL class ontology through the rdflib library; According to the labels in the RDF graph model, extracting entity data information of the OWL class ontology conforming to the element ontology structure and converting into a triple form; Traversing all OWL.Class labels in the RDF graph model, identifying and extracting rdfs:label labels under each Class entity to obtain basic information of the corresponding Class entity; adopting a three-level fallback analysis mechanism, identifying and extracting skos:prefLabel labels in the absence of rdfs:label labels, identifying and extracting URI fragments in the absence of skos:prefLabel labels, obtaining complete basic identification information of the Class entity, adopting a dynamic hierarchical calculation algorithm of cyclic detection, layer by layer accumulating inheritance paths of the concept class, deducing the hierarchy to which the Class entity belongs, and obtaining the triple form of the Class entity. Traverse all the owl.ObjectProperty labels in the RDF graph model, identify and extract the Class entities pointed by the rdfs:label label, rdfs:subPropertyOf label, rdfs:domain label and rdfs:range label of the ObjectProperty entity, construct the structured expression between the ObjectProperty entity and the Class entity as the head entity and the tail entity, and obtain the triple form of the ObjectProperty entity; Traverse all the owl.DatatypeProperty labels in the RDF graph model, identify and extract the information of the rdfs:label label, rdfs:domain label and rdfs:range label of the DataProperty entity, and the rdfs:range label corresponds to the data type of the DataProperty entity, and the rdfs:domain label corresponds to the associated Class entity, to obtain the triple form of the DataProperty entity; Step 3: Store the entity data information of the literature class ontology in the triple form and the entity data information of the OWL class ontology in the triple form into the ontology element knowledge graph to form the structured data layer of the ontology element knowledge graph; Step 4: Select the core element entities of the two types of ontologies in the structured data layer of the ontology element knowledge graph, vectorize the multiple features of the core element entities respectively to obtain the vectorized core element entity features, and calculate the similarity of the vectorized core element entity features of the literature class ontology and the vectorized core element entity features of the OWL class ontology to obtain the core element entity pair; The multiple features of the core element entity are the name feature representing the name of the core element entity, the definition structure feature representing the associated structure formed by the triple form, and the graph structure feature representing the structured relationship of the core element entity in the knowledge graph. The name feature comes from the head entity h or the tail entity t in the triple, the definition structure feature comes from the relationship r in the triple, and the graph structure feature is the entity and relationship network formed by multiple triples; Step 5: Traverse the core element entity pair, merge the core element entity pairs with similar mapping relationships, obtain the core element entity node and record the number of times the core element entity node is mapped, add the number of times the core element entity node is mapped to the authority value when the storage is completed, and obtain the final authority value of the core element entity node. 2.The ontology organization method based on a knowledge graph according to claim 1, characterized in that, In step 2, the knowledge extraction of the literature class ontology extracts the entity data information of the literature class ontology conforming to the element ontology structure and converts it into a triple form, including: Integrate the task target of literature class ontology knowledge extraction, the standard of knowledge extraction and the example of structured triple into a Prompt string, combine the Prompt string with the literature class ontology as input, call the large language model of ChatGPT-4o, use API as an interface, perform knowledge extraction on the literature class ontology, extract the entity data information of the literature class ontology conforming to the element ontology structure and convert it into a triple form. 3.The ontology organization method based on a knowledge graph according to claim 1, characterized in that, The carrier of the structured data layer of the ontology element knowledge graph in step 3 is a Neo4j graph database. 4.The ontology organization method based on a knowledge graph according to claim 1, wherein, The core element entities of the two types of ontologies in step 4 are Class entities, ObjectProperty entities and DataProperty entities. 5.The ontology organization method based on a knowledge graph according to claim 1, characterized in that, The name features and definition structure features are vectorized by using an all-MiniLM-L6-v2 model in a Sentence Transformer framework, and the graph structure features are vectorized by using a TransE model. 6.The ontology organization method based on a knowledge graph according to claim 1, wherein, In step 4, the core element entity features of the vectorized literature ontology and the core element entity features of the vectorized OWL ontology are subjected to similarity calculation to obtain core element entity pairs, including: In step 4, the core element entity features of the vectorized literature ontology and the core element entity features of the vectorized OWL ontology are subjected to similarity calculation to obtain core element entity pairs, including: 7.The ontology organization method based on a knowledge graph according to claim 1, characterized in that, In step 4, the core element entity features of the vectorized literature ontology and the core element entity features of the vectorized OWL ontology are subjected to similarity calculation to obtain core element entity pairs, including: 8.The ontology organization method based on a knowledge graph according to claim 1, wherein, In step 4, before the similarity calculation of the core element entity features of the vectorized literature ontology and the core element entity features of the vectorized OWL ontology, Cypher statements are used to remove core element entities that do not need to be traversed. In step 5, the number of times of mapping is added to the authoritative value at the time of storage to obtain the final authoritative value of the core element entity node, and the step 5 further includes: In step 3, during the storage of the entity data information into the ontology element knowledge graph, all entity data information in the form of triples is traversed, and the same core element entities are merged until the storage is completed, forming the structured data layer of the ontology element knowledge graph. The authoritative value of the core element entity is increased by one each time the core element entity is merged, until the authoritative value at the time of storage is obtained.

Images