A long text scientific data standardization semantic association publishing method and system

Abstract

The application relates to a long-text scientific data standardization semantic association publishing method and system. The method first carries out knowledge extraction on long-text scientific data to obtain semantic information; carries out data semantic standardization on the obtained semantic information, and constructs an RDF triple; and carries out data association publishing on the constructed RDF triple. The application faces long-text scientific data, adopts an integrated process, can directly publish long texts in the form of Linked Data, improves the semantic ability and association ability of long-text scientific data, carries out standardization semantic description on the extracted semantic information of the long texts, makes the long-text scientific data understood by the Web, is beneficial to association and fusion with other Linked Data open data, and makes scientific data resources described in the long texts more easily explored and discovered in the Web.

Description

Technical Field

[0001] This invention discloses a method and system for standardizing and publishing semantically related long-text scientific data, belonging to the field of semantic association technology. It can uniformly realize the complete process of knowledge extraction, data semantic standardization, and data association publishing from long-text scientific data. Background Technology

[0002] Linked data effectively addresses the issues of data heterogeneity and dissimilarity, enabling knowledge organization, association, and discovery among massive amounts of scientific data, thus serving as an effective means of knowledge sharing. With the promotion of the linked data concept and technological advancements, an increasing number of information repositories are adopting linked data formats for external publication and access services. Currently, a large amount of long-text data resources exist in scientific data, such as descriptive data on plant species morphology and geographical distribution in the plant science field, and soil type distribution data in the fields of glaciers, permafrost, and deserts. Long-text data contains rich semantic knowledge waiting to be mined. Publishing it in the form of linked data makes it understandable on the Web and facilitates the integration of multi-source data. Knowledge extraction from long-text data can utilize natural language processing techniques (such as the Language Technology Platform (LTP) of Harbin Institute of Technology and Stanford NLP of Stanford University) to extract semantic information in textual form. However, there are still many prominent problems in the standardized semantic linked publication of long-text data that urgently need to be addressed, specifically:

[0003] 1) Overall aspects:

[0004] (1) Long text knowledge extraction can only extract semantic information, lacks standardized semantic description, cannot be machine readable, and cannot be associated with and integrated with open resources in the semantic web.

[0005] (2) There is a lack of a complete process for extracting knowledge from long text scientific data and publishing it in the form of linked data.

[0006] 2) Knowledge extraction from long texts:

[0007] (1) Lack of standardized semantic description of named entities: Named entities lack semanticization and cannot be understood by the Web.

[0008] (2) Lack of deeper semantic mining: It is impossible to expand on the extracted natural language text again, and the degree of semantic mining is low.

[0009] 3) Data semantic standardization:

[0010] (1) Redundancy of synonymous data exists: Synonymous information exists in the extracted text, resulting in redundancy of published data.

[0011] (2) Unable to establish international links: Long text data cannot be linked with international Linked Data open data.

[0012] (3) Lack of standardized predicate description: The predicates extracted from long texts lack standardized descriptions, making them uncomprehensible to the Web.

[0013] (4) Lack of precise semantic description of objects: The objects extracted from long texts lack semantic description and cannot be understood by the Web. Summary of the Invention

[0014] To address the above-mentioned prominent issues, this invention proposes a method and system for standardized semantic association and publishing of long-text scientific data.

[0015] The technical solution adopted in this invention is as follows:

[0016] A method for standardizing and semantically associating long-text scientific data for publication includes the following steps:

[0017] Extract knowledge from long-text scientific data to obtain semantic information;

[0018] The acquired semantic information is standardized for data semantics, and RDF triples are constructed.

[0019] The constructed RDF triples are then used for data association and publication.

[0020] Furthermore, the knowledge extraction from long text scientific data to obtain semantic information includes:

[0021] Long scientific text data is broken down into several short texts using sentence segmentation methods in natural language processing. Then, each short text is processed using named entity recognition methods to identify the names of people, places, and organizations. The resource type is assigned using rdf:type.

[0022] Each short text is segmented, part-of-speech tagging is performed, and semantic dependency analysis is performed to obtain several words with semantic tags. The composition of each word in the sentence is determined based on different semantic tags. For each word with semantic tags, part-of-speech tagging, dependency parsing and other methods are used recursively to extract the subject-verb-object adjectives and combine them with the corresponding subject-verb-object.

[0023] The obtained objects containing adjectives are then subjected to part-of-speech tagging and dependency parsing to extract semantic information from the objects.

[0024] Furthermore, the data semantic standardization includes:

[0025] Thesaurus normalization: Select a thesaurus within the domain and map synonyms and non-technical terms in semantic information to preferred words in the thesaurus.

[0026] Semantic subject representation: Based on the Linked Data principle, create a web-accessible URI for the subject;

[0027] Predicate semantic standardization: Align the predicate with the standardized terms in the ontology, and construct an ontology to supplement the existing ontology;

[0028] Semantic object representation: Creates Web-understandable URIs for objects and associates them with Linked Data open resources.

[0029] Furthermore, the predicate semantic standardization includes: matching the predicate in text form with the predicate in the international ontology platform; aligning the matched predicates with standardized terms; and generating an ontology using RDF-related tools and adding attributes to the unmatched predicates, aligning the text predicates with the standardized terms in the ontology. The object semanticization includes: for objects not found in the semantic web, creating a web-accessible URI and recording the part of speech and corresponding predicate to improve matching accuracy.

[0030] Furthermore, the data semantic standardization uses URIs to establish associations with Linked Data open resources in international knowledge bases, thereby enhancing the internationalization of data association capabilities. In the process of association with international knowledge bases, local indexes are built for commonly used standardized attributes to improve knowledge base retrieval efficiency.

[0031] Furthermore, in the process of data semantic standardization, the international knowledge base is customizable. Users can select the required knowledge base or choose a knowledge base they create themselves. The knowledge base can be customized by setting SPARQL Endpoints for multiple knowledge bases.

[0032] Furthermore, the data association and publication of the constructed RDF triples is carried out through SPARQL Endpoint to provide retrieval services.

[0033] A standardized semantic association publishing system for long-text scientific data, comprising:

[0034] The long text knowledge extraction module is used to extract knowledge from long text scientific data and obtain semantic information from it.

[0035] The data semantic standardization module is used to standardize the acquired semantic information and construct RDF triples.

[0036] The association publishing module is used to associate and publish the constructed RDF triples.

[0037] The beneficial effects of this invention are as follows:

[0038] This invention targets long-text scientific data and employs an integrated workflow to directly publish long texts as Linked Data, enhancing their semantic and associative capabilities. It standardizes the semantic information extracted from long texts, enabling the Web to understand the data and facilitating its integration with other Linked Data open datasets. This makes the scientific data resources described in long texts easier to explore and discover on the Web.

[0039] This invention utilizes natural language processing techniques such as word segmentation, sentence segmentation, part-of-speech tagging, named entity recognition, and semantic dependency analysis, combined with a recursive approach, to extract semantic information in the form of <subject, predicate, object> triples from long text data. By mapping to a domain-specific thesaurus and aligning and associating with international ontology platforms, international knowledge bases, and domain-specific knowledge platforms, the semantic information is standardized and described. A standardized RDF triple model is then linked to the international knowledge base. The association and publication are facilitated through SPARQL endpoints, enabling semantic interoperability across multiple data sources. Attached Figure Description

[0040] Figure 1 This is the overall functional logic framework diagram of the present invention.

[0041] Figure 2 This is a flowchart illustrating the overall method steps of the present invention.

[0042] Figure 3 This is the result of knowledge extraction from long text.

[0043] Figure 4 This is the result of normalization processing of the thesaurus.

[0044] Figure 5 This is a flowchart of subject semanticization.

[0045] Figure 6 This is a flowchart of predicate semantic standardization.

[0046] Figure 7 This is the result of object semantic standardization.

[0047] Figure 8 It is the result of the long text scientific data semantic standardization program execution.

[0048] Figure 9 This is a SPARQL Endpoint query result for long text scientific data. Detailed Implementation

[0049] The present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0050] The overall innovations of this invention include:

[0051] 1. A method is proposed to publish long text scientific data in the form of Linked Data to the Semantic Web. By inputting long text scientific data, the semantic information in it can be published in the form of Linked Data, thereby improving the semantic and associative capabilities of long text scientific data.

[0052] 2. Expand the original content of long-text scientific data, and extend the data relevance based on the existing semantics.

[0053] The overall logic of this invention is as follows: Figure 1 As shown: Input long-text scientific data and use natural language processing tools to extract semantic information from the long text. Select a thesaurus within the domain, and normalize the extracted semantic information according to the thesaurus. Map synonyms in the triples (subject, predicate, object) and non-technical terms to preferred words in the thesaurus. During semantic normalization, according to Linked Data principles, create a Web-accessible URI for the subject, align the predicate with standardized terms in the ontology, and construct an ontology to supplement the existing ontology. Create a Web-understandable URI for the object and associate it with Linked Data open resources. Finally, construct RDF (Resource Description Framework) triples for data association and publication.

[0054] The overall methodology can be summarized into three main steps, such as... Figure 2 As shown, the meaning of each step is explained below:

[0055] 1. Knowledge Extraction from Long Texts: Using natural language processing techniques such as word segmentation, sentence segmentation, named entity recognition, dependency parsing, and semantic dependency parsing to extract semantic information from long texts.

[0056] 2. Data Semantic Standardization: Select a thesaurus of scientific data within the domain and map semantic information to the preferred terms in the thesaurus. Then, match it with international knowledge bases and ontology platforms, and convert the semantic information into RDF format based on the Linked Data principle to make it machine-readable.

[0057] 3. Related Data Publishing: Related data publishing is performed through SPARQL Endpoints to provide search services.

[0058] The design of each step is explained in detail below:

[0059] 1. Knowledge Extraction from Long Texts

[0060] The long text scientific data was decomposed into several short texts using sentence segmentation methods in natural language processing. Then, each short text was processed using named entity recognition to identify the names of people, places, and organizations. The resource type was assigned using rdf:type (see Innovation Points of Long Text Knowledge Extraction (1)). Each short text was segmented, part-of-speech tagging was performed, and semantic dependency analysis was performed to obtain several words with semantic tags. The composition of each word in the sentence was determined based on the different semantic tags. For each word with semantic tags, part-of-speech tagging, dependency parsing, and other methods were recursively used to extract the adjectives of subject, verb, and object, and combined with the corresponding subject, verb, and object. The obtained objects with adjectives were then subjected to part-of-speech tagging, dependency parsing, and other methods to extract the semantic information in the objects (see Innovation Points of Long Text Knowledge Extraction (2)). The results are as follows. Figure 3 As shown, this allows for the extraction of more knowledge from long texts and enables greater connections with open knowledge bases.

[0061] The following details the innovations of this invention from the perspective of knowledge extraction from long texts:

[0062] (1) Named Entity Standardized Semantic Description: Named entities identified after extraction are described using standardized semantics to make them understandable on the Web. Named entity recognition methods are used to identify person names, place names, and organization names in long texts. The resource type is assigned using `rdf:type`, where `rdf:type` represents a term from the RDF dictionary used to express the resource type. The complete URI is: https: / / www.w3.org / 1999 / 02 / 22-rdf-syntax-ns#type. For example, a person name is assigned the type `foaf:Person`, where `foaf:Person` represents a term from the `foaf` dictionary used to express the person class. The complete URI is:

[0063] http: / / xmlns.com / foaf / 0.1 / Person. This could make semantic markup more complete.

[0064] (2) Deeper semantic mining: The extracted natural language text is further expanded to mine deeper semantic relationships. Continue to mine the semantic information of objects extracted from long texts so that more semantic information can be associated with open data.

[0065] 2. Data semantic standardization

[0066] First, the semantic information extracted from the long text is normalized by referring to the thesaurus (see Data Semantic Standardization Innovation Points (1)). The thesaurus is a standardized dictionary that summarizes a specific knowledge domain. It consists of semantically related attributes and can be continuously supplemented and modified. The thesaurus is maintained by the user and stored in a specific format, such as database, EXCEL, text, etc., as shown in Table 1. There may be synonyms in the semantic information extracted from the long text, but some words are not professional terms in this field. At this time, the thesaurus is used for normalization. The thesaurus in this field is selected, and the subject, verb, and object of each text form are queried in the thesaurus. The synonyms and non-professional terms in the semantic information are replaced with the preferred words in the thesaurus. The result of the thesaurus normalization is as follows. Figure 4 As shown.

[0067] Table 1. Synonyms for the Plant Domain

[0068]

[0069] 2.1 Subject Semanticization

[0070] Semantic subject representation aims to improve relevance with international knowledge bases and requires matching with data in international open knowledge bases. Currently, mainstream international knowledge bases include DBpedia, Yago, and subject-specific knowledge bases such as Bio2RDF and Openbiodiv. Users can add knowledge bases from different fields through configuration (see Semantic Standardization Innovation Point (2)). However, international knowledge bases contain a huge amount of data, resulting in slow query speeds and low efficiency. By building indexes through their open SPARQL Endpoints (see Semantic Standardization Innovation Point (3)), matching efficiency can be improved.

[0071] This article adds the owl:sameAs relationship to subjects matched in the knowledge base index, associates them with international knowledge bases, and generates web-accessible URIs for them. Subjects that do not match are directly assigned web-accessible URIs. Here, owl:sameAs represents a term from the owl vocabulary used to express the equality relationship between two resources; the complete URI is...

[0072] http: / / www.w3.org / 2002 / 07 / owl#sameAs.

[0073] Taking plant-related data as an example, based on the Linked Data principle, URIs are used to identify the subjects extracted from long texts, and these URIs are made accessible on the Web. The subject-assigned URI format is as follows:

[0074] <baseuri> / <name>

[0075] The base URL (baseUri) is set to http: / / www.example.com, and the subject name (name) is used as the second-level URL. This allows for the association of the same subject across multiple text segments and can also be combined with related data published in any other form, improving data discoverability and facilitating association with other open resources. The subject semanticization process is as follows: Figure 5 As shown.

[0076] 2.2 Predicate Semantic Standardization

[0077] The predicate semantic standardization process requires matching with international ontology platforms, such as EBI-OLS and Ontobee. Users can also add domain-specific ontology based on their configurations (see Semantic Standardization Innovation Point (2)). However, international ontology platforms contain massive amounts of data, making direct matching slow and inefficient. Indexes can be built using the open APIs of international ontology platforms (see Semantic Standardization Innovation Point (3)) to improve retrieval efficiency.

[0078] The predicates after the thesaurus normalization are still in text form. Predicates are matched in the index, and those that match are represented using standardized ontology attributes (see Semantic Standardization Innovation Points (4)). An ontology can also be constructed based on long text content, mapping the predicates. For predicates that cannot be matched, RDF processing techniques can be used, such as Apache Jena and RDF4j, to construct the ontology and add attributes. Returning the ontology used and created during the semantic standardization process facilitates the updating and improvement of ontology within the domain, and also makes it convenient for other data to use the ontology for standardization. The predicate semantic standardization process is as follows: Figure 6 As shown.

[0079] 2.3 Object Semantics

[0080] The object can be a resource or a literal. If the object is a resource entity, it is matched in the knowledge base index constructed in 2.1, and the returned URI is used to replace the text type object (see Semantic Standardization Innovation Point (5)). These object URIs are associated with existing resources in the international knowledge base using attributes such as owl:sameAs, skos:exactMatch, and rdfs:subClassOf. Among them, skos:exactMatch represents a term in the skos thesaurus, which is used to express that the resource has an exact match. The complete URI is: http: / / www.w3.org / 2004 / 02 / skos / core#exactMatch, and rdfs:subClassOf represents a term in the rdfs thesaurus, which is used to express that the current class is a subclass of another class. For object resources that cannot be associated and discovered in the index, a URI is assigned to make them accessible to the Web, and the part of speech and corresponding predicate of the object are recorded (see Semantic Standardization Innovation Point (6)) to realize the reuse of URI so that new data can be matched with it. If the object is a literal, XML Schema is used to describe the datatype. The semantic object rendering process is as follows: Figure 7 As shown.

[0081] The following section details the innovations of this invention from the perspective of data semantic standardization:

[0082] (1) Synonym normalization: For synonyms appearing in the text, a thesaurus is constructed, and the preferred word is used to replace the synonyms to improve the standardization of the extracted data.

[0083] (2) Customizable international knowledge base: Different fields involve different knowledge bases. Users can choose the required knowledge base or choose their own created knowledge base. Multiple SPARQL Endpoints can be set to realize the customization of knowledge bases.

[0084] (3) Improved efficiency of international association: Association with international knowledge bases requires calling their open APIs for queries. However, international knowledge bases contain massive amounts of data, making direct queries slow and inefficient. Building local indexes for commonly used standardized attributes can improve knowledge base retrieval efficiency.

[0085] (4) Standardized description of predicates: The predicates in text form are matched with the predicates in the international ontology platform. If they can be matched, they are aligned using standardized terms. If they cannot be matched, RDF-related tools are used to generate an ontology and add attributes. Aligning the predicates in text form with the standardized terms in the ontology improves the standardization of resource descriptions and makes them easier to understand in the Web.

[0086] (5) Establishing international connections: If the semantic information extracted from long text scientific data can be matched in the international knowledge base, the existing URI will be used to represent it and establish a connection with the Linked Data open resources in the international knowledge base to improve the international connection capability of the data.

[0087] (6) Precise semanticization of objects: The same object may have different meanings when combined with different predicates. For objects not found in the semantic web, create web-accessible URIs and record the part of speech and corresponding predicates to improve matching accuracy.

[0088] 3. Data Association and Publishing

[0089] After subject-verb-object standardization, the standardized semantic information described in RDF triples is understandable to the web, such as... Figure 8 As shown, the RDF model is uploaded to an RDF database and published via a SPARQL endpoint to provide retrieval services. Based on the linked data principle, when accessing the URI, the Sparql endpoint is queried, and Jena is used to parse the relevant triples to provide the RDF data. The RDF database can be any tool such as Fuseki or Virtuoso. The data retrieved through Sparql statements is as follows: Figure 9 As shown.

[0090] Through the above steps, standardized semantic publishing of long-text scientific data has been achieved, which has enabled deep integration with international open data, improved the machine readability of long-text scientific data, and facilitated further exploration of semantic relationships at the data content level.

[0091] 4. Summary of the Invention

[0092] This invention proposes a standardized semantic association method for publishing long-text scientific data. Employing an integrated process, it can directly publish long texts as Linked Data, and standardizes the semantic description of the extracted data, thereby improving the semantic and association capabilities of long-text scientific data. This overcomes the current limitations of directly publishing long-text scientific data as linked data and its inability to be understood by the Web, and solves the problem of weak association with open data. Possible innovations of this invention include:

[0093] 1) Overall aspects:

[0094] (1) A method is proposed to publish long text scientific data in the form of Linked Data to the Semantic Web.

[0095] (2) Expand the original content of long text scientific data and expand the data relevance based on the existing semantics.

[0096] 2) Knowledge extraction from long texts:

[0097] (1) Standardized semantic description of named entities: For the named entities identified after extraction, standardized semantics are used to describe them so that the Web can understand them.

[0098] (2) Deeper semantic mining: The extracted natural language text is further expanded to mine deeper semantic relationships.

[0099] 3) Data semantic standardization:

[0100] (1) Synonym unification: For synonyms appearing in the text, a thesaurus is constructed to improve the standardization of the extracted data.

[0101] (2) Customizable international knowledge base: Multiple SPARQL Endpoints for knowledge bases can be set to enable customizable knowledge bases.

[0102] (3) Improved efficiency of internationalization: For commonly used standardized attributes, local indexes are built to improve the efficiency of knowledge base retrieval.

[0103] (4) Standardized description of predicates: Align the predicates in text form with the standardized terms in the ontology to improve the standardization of resource descriptions and make them easier to understand in the Web.

[0104] (5) Establishing international connections: Establish connections with Linked Data open resources in international knowledge bases to enhance the ability to connect data internationally.

[0105] (6) Precise semanticization of objects: For objects not found in the semantic web, create web-accessible URIs and record the part of speech and corresponding predicates to improve matching accuracy.

[0106] Another embodiment of the present invention provides a long text scientific data standardization semantic association publishing system, which includes:

[0107] The long text knowledge extraction module is used to extract knowledge from long text scientific data and obtain semantic information from it.

[0108] The data semantic standardization module is used to standardize the acquired semantic information and construct RDF triples.

[0109] The association publishing module is used to associate and publish the constructed RDF triples.

[0110] For the specific implementation process of each module, please refer to the description of the method of the present invention above.

[0111] Based on the same inventive concept, another embodiment of the present invention provides a computer device (computer, server, smartphone, etc.) including a memory and a processor, wherein the memory stores a computer program configured to be executed by the processor, and the computer program includes instructions for performing the steps of the method of the present invention.

[0112] Based on the same inventive concept, another embodiment of the present invention provides a computer-readable storage medium (such as ROM / RAM, disk, optical disk), which stores a computer program that, when executed by a computer, implements the various steps of the method of the present invention.

[0113] The specific embodiments of the present invention disclosed above are intended to help understand the content of the present invention and to implement it accordingly. Those skilled in the art will understand that various substitutions, changes, and modifications are possible without departing from the spirit and scope of the present invention. The present invention should not be limited to the content disclosed in the embodiments of this specification; the scope of protection of the present invention is defined by the claims.< / name> < / baseuri>

Claims

1. A method for standardized semantic association and publishing of long-text scientific data, characterized in that, Includes the following steps: Extract knowledge from long-text scientific data to obtain semantic information; The acquired semantic information is standardized for data semantics, and RDF triples are constructed. The constructed RDF triples are then linked and published. The data semantic standardization includes: Thesaurus normalization: Select a thesaurus within the domain and map synonyms and non-technical terms in semantic information to the preferred words in the thesaurus; Subject semanticization: Based on the Linked Data principle, create web-accessible URIs for subjects; add owl:sameAs relationships to subjects matched in the knowledge base index, associate them with international knowledge bases, and generate web-accessible URIs for them; for subjects that cannot be matched, directly generate web-accessible URIs. Here, owl:sameAs represents a term in the owl vocabulary, used to express the equality relationship between two resources. Predicate semantic standardization: Align the predicate with the standardized terms in the ontology, and construct an ontology to supplement the existing ontology; match the predicate in text form with the predicate in the international ontology platform. If a match can be made, use standardized terms to align it. If a match cannot be made, use RDF-related tools to generate an ontology and add attributes to align the predicate in text form with the standardized terms in the ontology. Semantic object representation: Create Web-understandable URIs for objects and associate them with Linked Data open resources; if the object is a resource entity, match it in the constructed knowledge base index, replace the text-type object with the returned URI, and establish association between the object URI and existing resources in the international knowledge base using the attributes owl:sameAs, skos:exactMatch, and rdfs:subClassOf. Here, skos:exactMatch represents a term from the skos thesaurus to indicate an exact match, and rdfs:subClassOf represents a term from the rdfs thesaurus to indicate that the current class is a subclass of another class; for object resources that cannot be found in the index, assign a URI to make them Web-accessible, and record the part-of-speech and corresponding predicate of the object to achieve URI reuse; if the object is a literal, use XML Schema to describe the data type.

2. The method according to claim 1, characterized in that, The process of extracting knowledge from long-text scientific data to obtain semantic information includes: Long scientific text data is broken down into several short texts using sentence segmentation methods in natural language processing. Then, each short text is processed using named entity recognition methods to identify the names of people, places, and organizations. The resource type is assigned using rdf:type. Each short text is segmented, part-of-speech tagging is performed, and semantic dependency analysis is performed to obtain several words with semantic tags. The composition of each word in the sentence is determined based on different semantic tags. For each word with semantic tags, part-of-speech tagging, dependency parsing and other methods are used recursively to extract the subject-verb-object adjectives and combine them with the corresponding subject-verb-object. The obtained objects containing adjectives are then subjected to part-of-speech tagging and dependency parsing to extract semantic information from the objects.

3. The method according to claim 1, characterized in that, The data semantic standardization uses URIs to establish associations with Linked Data open resources in international knowledge bases, thereby enhancing the internationalization of data association capabilities. In the process of association with international knowledge bases, local indexes are built for commonly used standardized attributes to improve knowledge base retrieval efficiency.

4. The method according to claim 1, characterized in that, During the data semantic standardization process, the international knowledge base is customizable. Users can choose the required knowledge base or choose a knowledge base they create themselves. The customizability of the knowledge base is achieved by setting SPARQL Endpoints for multiple knowledge bases.

5. The method according to claim 1, characterized in that, The process of publishing the constructed RDF triples is achieved through SPARQL Endpoints, providing retrieval services.

6. A long-text scientific data standardization and semantic association publishing system employing the method described in any one of claims 1 to 5, characterized in that, include: The long text knowledge extraction module is used to extract knowledge from long text scientific data and obtain semantic information from it. The data semantic standardization module is used to standardize the acquired semantic information and construct RDF triples. The association publishing module is used to associate and publish the constructed RDF triples.

7. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program configured to be executed by the processor, the computer program including instructions for performing the method of any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, which, when executed by a computer, implements the method according to any one of claims 1 to 5.

Images