A Method for Constructing a Knowledge Graph for Risk Assessment of Immovable Cultural Relics

Abstract

The present invention discloses a method for constructing a knowledge graph for the risk assessment of immovable cultural relics, including ontology model design, multi-source heterogeneous data processing, a mapping method from RDF data in the field of immovable cultural relics risk assessment to a graph database, and a method for constructing a domain knowledge graph based on the Neo4J graph database; the ontology of immovable cultural relics and risk assessment information are stored in the form of a graph database, the seven-step method, which is a domain ontology modeling method for the risk assessment of immovable cultural relics, is improved, and an ontology model is established according to the improved method. After the ontology model is constructed, a knowledge graph for the risk assessment of immovable cultural relics is constructed using the graph database Neo4J based on the ontology model and domain data.

Description

Technical Field

[0001] The present invention relates to the research field of constructing domain knowledge graphs, and particularly to a method for constructing a knowledge graph in the field of immovable cultural relics risk assessment. Background Art

[0002] The multi-source heterogeneous data involved in immovable cultural relics risk assessment has the characteristics of various types, wide distribution range, large data volume, and multiple dimensions such as time and space. These characteristics lead to an increase in the complexity of data management. Therefore, an integration method is needed to organize the scattered data and describe the risk factors in all aspects of immovable cultural relics. When conducting risk analysis, it can provide efficient and accurate data support for immovable cultural relics risk assessment. With the proposal of knowledge graphs, many data integration platforms have begun to apply knowledge graphs to connect the originally unconnected data and integrate the discrete data together, so as to provide more valuable decision-making support. Therefore, for the above reasons, the present invention proposes a method for constructing a knowledge graph for immovable cultural relics risk assessment for immovable cultural relics risk assessment

[0003] The concept of Knowledge Graph was officially proposed by Google in 2012 with the aim of achieving a more intelligent search engine. Essentially, a knowledge graph is a knowledge base called a Semantic Network, that is, a knowledge base with a directed graph structure, where the nodes of the graph represent entities or concepts, and the edges represent various semantic relationships between entities / concepts. Knowledge graphs are divided into general knowledge graphs and domain knowledge graphs. General knowledge graphs are oriented towards general domains and mainly contain a large amount of common sense knowledge in the real world, with a wide coverage. Domain knowledge graphs are oriented towards a specific domain and are industry knowledge bases composed of professional data in that domain, with high requirements for the depth and accuracy of domain knowledge. Domain knowledge graphs have many different data models to adapt to different business scenarios and users.

[0004] The methods for constructing domain knowledge graphs mainly include three types: top-down, bottom-up, and hybrid. The top-down construction form is to first extract ontology and schema information from high-quality data with the help of structured data sources such as encyclopedic websites at the initial stage of constructing the knowledge graph and add them to the knowledge base. The bottom-up construction form applies knowledge extraction technology to continuously extract and update concepts and the organizational structure between concepts from data sources without a defined organizational structure. The hybrid construction method of top-down and bottom-up has a predefined ontology or schema layer at the initial stage, and at the same time, as data sources are added, the organizational structure of the data sources can update and improve the knowledge graph schema layer, thereby improving the integrity and reliability of the knowledge graph. Summary of the Invention

[0005] In view of the problems existing in the risk assessment of immovable cultural relics, such as the variety of cultural relic risk data, complex relationships, and the inconvenience for risk assessment experts to evaluate cultural relic risks, the present invention proposes a method for constructing a knowledge graph for the risk assessment of immovable cultural relics. Based on the domain knowledge of the risk of immovable cultural relics, this method stores the ontology of immovable cultural relics and risk assessment information in the form of a graph database, improves the domain ontology modeling method - the seven-step method according to the risk assessment model of immovable cultural relics, and establishes an ontology model according to the improved method. After the ontology model is constructed, a knowledge graph for the risk assessment of immovable cultural relics is constructed using the graph database Neo4J according to the ontology model and domain data.

[0006] The method proposed by the present invention mainly includes ontology model design, multi-source heterogeneous data processing, the mapping method from RDF data in the field of immovable cultural relic risk assessment to a graph database, and the construction method of a domain knowledge graph based on the Neo4J graph database, which are introduced in detail as follows:

[0007] Ontology model design: The ontology model is the conceptual model of the knowledge graph, providing the data organization structure for the construction of the knowledge graph in the field of immovable cultural relic risk assessment. By analyzing the seven-step method in detail and combining the characteristics of the immovable cultural relic risk assessment field, the present invention adds the processes of ontology evaluation and iteration to the seven-step method to improve the defects of ignoring content update and lacking scientific management and evaluation mechanisms in the seven-step method. In addition, in the process of ontology construction, the present invention refers to the risk assessment model of immovable cultural relics given by domain experts in the project team for ontology design, thereby reducing the problem of strong subjectivity. The immovable cultural relic risk assessment ontology model has a total of eight steps, namely, clarifying the domain scope, examining the possibility of reusing existing ontologies, listing important terms in the ontology, defining the hierarchical relationship between classes, defining the attributes of classes and attribute constraints, ontology instantiation, ontology evaluation, ontology establishment, and iteration. The present invention designs and implements the immovable cultural relic risk assessment ontology model through the above steps and uses the Protégé ontology modeling tool. Finally, the ontology model is saved in RDF format.

[0008] Multi-source heterogeneous data processing: The data sources involved in constructing the knowledge graph for the risk assessment of immovable cultural relics mainly include: network data, structured data, text data, and field-collected data. Most of the data obtained from these data sources is in a relatively chaotic format and does not conform to the specifications set by the system. Therefore, means such as regular matching, web parsing, and data normalization are required to further extract and transform the data. For text data, relevant data contained in the text data is extracted through keyword extraction, regular matching, and manual methods, and saved in the form of an Excel table; network data mainly includes websites containing immovable cultural relic data such as the official website of the Chinese Academy of Cultural Heritage. The content of these websites is extracted through web parsing technology, and the obtained content is stored in the form of an Excel table; for structured data, we perform standardized processing according to the data standardization specifications provided by experts in the field of risk assessment of immovable cultural relics. Through the above steps, we finally obtain the original data for constructing the knowledge graph for the risk assessment of immovable cultural relics.

[0009] Mapping method of RDF data in the field of risk assessment of immovable cultural relics to a graph database: This step aims to map the RDF data generated by the ontology model for the risk assessment of immovable cultural relics into the format of an attribute graph that can be stored in the Neo4J graph database. After establishing the ontology model for the risk assessment of immovable cultural relics, it is usually stored using an RDF file. RDF data uses triples <subject, predicate, object> to represent the relationships between nodes in the knowledge graph for the risk assessment of immovable cultural relics, thus forming the semantic network structure for the risk assessment of immovable cultural relics.

[0010] r = <s, p, o>

[0011] Among them, s represents the subject of the immovable cultural relic network structure, p represents the predicate of the immovable cultural relic network structure, and o represents the object of the immovable cultural relic network structure.

[0012] Neo4J stores data in the form of an attribute graph. An attribute graph consists of a set of nodes, a set of relationships, a set of attribute key-value pairs, a path start-end mapping function, and labels.

[0013] G = <V, E, P, src, tgt, lbl>

[0014] Among them, V is the set of nodes in the graph database; E is the set of edges; P is the set of attribute key-value pairs; src, tgt, and lbl all represent functional relationships, indicating that each edge in the graph has a start point, each edge in the graph has an end point, and each node / edge in the graph model can be set with zero or more labels to identify.

[0015] 1) Node mapping

[0016] Let \(S\) be the set of subjects of RDF and \(O\) be the set of objects of RDF. Map \(S\) and \(O\) to \(V\) such that there are no duplicate elements in \(V\). Denote the set of classes in the RDF file as \(C\), and map the class \(c (c\in C)\) to which each element in \(S\) and \(O\) in RDF belongs to the label \(lbl\) of the corresponding element in \(V\).

[0017] 2) Relationship mapping

[0018] The relationship between two classes in the ontology actually reflects the possible relationships between the instances belonging to these two classes. Therefore, it can be mapped to the relationship in the Neo4j graph database. Since the classes in the ontology correspond to the labels in Neo4j, the relationships in the ontology can also be mapped to the possible relationships between the nodes of these two labels.

[0019] 3) Property mapping

[0020] In the ontology model, data properties describe the characteristics of classes in the ontology model. It can be directly corresponding to the property key-value pair \(P\) of the Neo4j node.

[0021] Method for constructing a domain knowledge graph based on the Neo4J graph database: According to the above mapping rules from the ontology to Neo4j, the present invention organizes and stores the immovable cultural relic risk assessment data into the Neo4J graph database according to the immovable cultural relic risk assessment ontology.

[0022] Compared with the prior art, the present invention has the following characteristics:

[0023] 1. Introduce the knowledge graph into the field of immovable cultural relic risk assessment, provide convenient and efficient knowledge services for the risk assessment work, and improve the assessment efficiency.

[0024] 2. The ontology model construction method is improved on the basis of the seven-step method, adding the process of ontology evaluation and iteration to improve the defects of ignoring content update and lacking scientific management and evaluation mechanism in the seven-step method. And refer to the opinions of domain experts during the ontology construction process, thereby reducing the problem with strong subjectivity.

[0025] 3. Use Neo4J to realize the storage and visualization of the knowledge graph, and provide an intuitive and friendly visualization interface for non-professionals. Brief description of the drawings

[0026] Figure 1 It is the process of the domain ontology modeling method improved by the present invention on the basis of the seven-step method.

[0027] Figure 2 It is the relationship diagram of the immovable cultural relic risk assessment ontology model.

[0028] Figure 3 It is the flowchart of multi-source heterogeneous data processing.

[0029] Figure 4 This is an example of the visualization effect of the knowledge graph in the field of immovable cultural relics risk assessment. Specific implementation manner

[0030] The present invention proposes a method for constructing a knowledge graph for immovable cultural relics risk assessment. The method mainly includes ontology model design, multi-source heterogeneous data processing, a mapping method from RDF data in the field of immovable cultural relics risk assessment to a graph database, and a method for constructing a domain knowledge graph based on the Neo4J graph database.

[0031] The following describes in detail the method for constructing a knowledge graph for immovable cultural relics risk assessment proposed by the present invention in combination with specific implementations.

[0032] First is the ontology model design. The present invention analyzes the seven-step method in detail and combines the characteristics of the immovable cultural relics risk assessment field. On the basis of the seven-step method, the ontology evaluation and iteration processes are added to improve the defects of the seven-step method that ignores content updates and lacks a scientific management and evaluation mechanism. In addition, in the process of ontology construction, the present invention refers to the immovable cultural relics risk assessment model given by domain experts in the project team to design the ontology, thereby reducing the problem of strong subjectivity. The present invention combines the above ideas to form a set of ontology modeling methods for the immovable cultural relics risk field. The specific process is as Figure 1 shown.

[0033] 1) Define the domain scope

[0034] The ontology in the field of immovable cultural relics risk assessment is used to describe various risk indicators involved in the immovable cultural relics ontology and the relationships between various risk indicators, providing effective support for the risk assessment of immovable cultural relics. Relevant domain information is obtained through channels such as books, literature, network materials, and consulting domain experts, and relevant concepts in the field of immovable cultural relics risk assessment are abstracted. The risk situation of immovable cultural relics is described from four aspects: the danger of disaster-causing factors, the vulnerability of the carrier, the sensitivity of the disaster-forming environment, and the disaster prevention and mitigation ability, and the relationships between relevant risk indicators are organized.

[0035] 2) Examine the possibility of reusing existing ontologies

[0036] Through a large number of research works, it is found that most of the existing cultural relics ontology models are constructed based on the information of cultural relics themselves and there is no ontology that can be used in the field of immovable cultural relics risk assessment. Therefore, it is impossible to reuse the existing ontology models in the process of constructing the ontology in the field of immovable cultural relics risk assessment. However, for the existing cultural relics ontology models constructed based on cultural relics ontology information, for example, the design of the architecture and mural ontologies is borrowed from the cultural relics ontology designed by Lin Yangping [9], and the ontology design of stone carving cultural relics is borrowed from the Southern Dynasties mausoleum stone carving ontology designed by Li Yonghui

[10] , etc.

[0037] 3) List important terms in the ontology

[0038] Obtain terms in the field of immovable cultural relics by referring to the existing cultural relic ontology model and official documents such as the industry guidelines issued by the National Cultural Heritage Administration and the draft for comments on the standard specifications for metadata for the digital protection of cultural relics; extract data-related terms by classifying and analyzing the data, and discuss the standardization and necessity of the terms with domain experts. The term extraction results are shown in Table 1.

[0039] Table 1 Terms in the field of risk assessment of immovable cultural relics (main)

[0040]

[0041] 4) Define the hierarchical relationship between classes

[0042] A class is an abstraction of things and can also be understood as a concept. Classes usually form a classification hierarchy (subclass-superclass hierarchy), and its hierarchical structure is usually an is-a hierarchy: a subclass is an instance of a superclass. By defining the hierarchical relationship between classes, the concepts in the field can be clarified. In Protégé, the Thing class is the superclass of all classes, and classes in the field of risk assessment of immovable cultural relics are defined under the Thing class. In the class design of the present invention, the ontology is divided into two parts. The first part describes immovable cultural relics and their basic information, including the class of immovable cultural relics, the class of regions, the class of cultural relic types, etc.; the second part describes risk index information, including the class of disaster-causing factor hazards of immovable cultural relics, the class of carrier vulnerabilities, the class of disaster-bearing environment sensitivities, the class of disaster prevention and mitigation capabilities, etc.

[0043] 5) Define the properties of classes and property constraints

[0044] Only defining the hierarchical relationship between classes cannot fully describe the knowledge in the field of risk assessment of immovable cultural relics, and it also needs to be improved by defining the properties of classes and property constraints. There are two types of properties in the ontology, namely object properties and data properties. Object properties connect two objects (subject and object) with a predicate, while data properties describe a certain object through a certain type of attribute data. Combine domain knowledge and risk assessment models to define the properties of classes and property constraints in the domain ontology.

[0045] 6) Ontology instantiation

[0046] The ultimate goal of building an ontology model is to apply the ontology model in practice to solve practical problems in the field. Therefore, it is necessary to combine domain data to instantiate the classes in the ontology of risk assessment of immovable cultural relics. Instantiating the ontology can not only enrich the content of the ontology but also evaluate the quality of the ontology using the instances.

[0047] 7) Ontology Evaluation

[0048] Ontology evaluation mainly includes two aspects: evaluation of ontology construction specifications and evaluation by domain experts. Ontology construction mainly needs to follow the following basic principles: clarity and objectivity, consistency, scalability, minimum coding bias, and minimum ontology commitment. When evaluating ontology construction specifications, it is necessary to evaluate the ontology quality according to the above basic principles. Domain expert evaluation is to evaluate whether the ontology design conforms to industry specifications and meets the needs of problem-solving under the guidance of domain experts after ontology instantiation.

[0049] 8) Ontology Establishment and Iteration

[0050] Finally, an ontology model for the risk assessment of immovable cultural relics that meets the requirements of ontology evaluation is obtained according to the above steps and applied to the solution of practical problems. If new terms are added or the ontology needs to be modified during use, ontology iteration is required.

[0051] According to the above method steps, the ontology model for the risk assessment of immovable cultural relics has been constructed using the Protégé tool. The relationships between ontology classes are as Figure 2 shown. After the ontology construction is completed, the ontology is exported in the RDF file format for use in subsequent steps.

[0052] Secondly, multi-source heterogeneous data is processed to obtain the original data for constructing the knowledge graph of the risk assessment of immovable cultural relics. The specific process is as Figure 3 shown. The data sources involved in constructing the knowledge graph mainly include: network data, structured data, text data, and field-collected data. Most of the data obtained from these data sources is in a relatively chaotic format and does not conform to the system-set specifications. Therefore, it is necessary to further extract and transform the data to finally obtain the original data for constructing the knowledge graph of the risk assessment of immovable cultural relics. Among them, for text data, relevant data contained in the text data is extracted through keyword extraction, regular matching, and manual methods and saved in the form of an Excel table; network data mainly includes websites such as the official website of the Chinese Academy of Cultural Heritage that contain data on immovable cultural relics. The content of these websites is extracted through web parsing technology and stored in the form of an Excel table; generally, structured data and field-collected data are already stored in the form of tables, so this type of data only needs to be standardized. After the above heterogeneous data is extracted into structured data, data extraction and transformation are carried out according to the risk data extraction rules provided by domain experts, so as to obtain the original data for constructing the knowledge graph of the risk assessment of immovable cultural relics.

[0053] Finally, it is the mapping of RDF data in the field of immovable cultural relics risk assessment to the graph database and the construction of the domain knowledge graph based on the Neo4J graph database. According to the mapping rules of RDF data to the graph database introduced in the invention content, the immovable cultural relics risk assessment data is organized according to the immovable cultural relics risk assessment ontology and stored in the Neo4J graph database. First, read the RDF file storing the immovable cultural relics risk assessment ontology through the JaneAPI to obtain all the triples that make up the ontology model. Then read the immovable cultural relics risk assessment data and organize the data according to the mapping rules. Finally, connect to the graph database through JDBC to execute the construction command to complete the storage of the data.

[0054] After the construction of the knowledge graph is completed, the node relationship diagram of the knowledge graph can be viewed on the Web page. At present, only the risk data of immovable cultural relics in Shanxi Province is constructed according to the method of the present invention. The knowledge graph includes 170,807 entities (including 3,208 immovable cultural relics entities) and 174,135 relationships. Figure 4 It shows a partial view of the knowledge graph. From the figure, it can be clearly seen the relationship between the index data related to various risks of immovable cultural relics and each index data.

Claims

1. A method for constructing a knowledge graph for the risk assessment of immovable cultural relics, characterized in that, It includes the ontology model design, the processing of multi-source heterogeneous data, the mapping method of RDF data in the field of immovable cultural relics risk assessment to the graph database, and the construction method of the domain knowledge graph based on the Neo4J graph database. The specific implementation process is as follows. Ontology model design: The ontology model is the conceptual model of the knowledge graph, providing the organizational structure of data for the construction of the knowledge graph in the field of immovable cultural relics risk assessment. By analyzing the seven-step method in detail and combining the characteristics of the immovable cultural relics risk assessment field, refer to the immovable cultural relics risk assessment model given by domain experts in the project team during the ontology construction process for ontology design. Use the Protégé ontology modeling tool to design and implement the immovable cultural relics risk assessment ontology model, and save the ontology model in RDF format. Multi-source heterogeneous data processing: The data sources involved in constructing the immovable cultural relics risk assessment knowledge graph include: network data, structured data, text data, and field-collected data. Use regular matching, web parsing, and data normalization to extract and transform the data to obtain the original data for constructing the immovable cultural relics risk assessment knowledge graph. Mapping method of RDF data in the field of immovable cultural relics risk assessment to the graph database: Map the RDF data generated by the immovable cultural relics risk assessment ontology model into the format of the property graph that can be stored in the Neo4J graph database. After establishing the immovable cultural relics risk assessment ontology model, use the RDF file for storage. The RDF data uses triples <subject, predicate, object> to represent the relationships between nodes in the immovable cultural relics risk assessment knowledge graph, thus constituting the semantic network structure of immovable cultural relics risk assessment. r = <s, p, o> Where s represents the subject of the immovable cultural relics network structure, p represents the predicate of the immovable cultural relics network structure, and o represents the object of the immovable cultural relics network structure. Neo4J stores data in the form of a property graph. The property graph consists of a node set, a relationship set, a property key-value pair set, a path start-end mapping function, and labels. G = <V, E, P, src, tgt, lbl> Where V is the node set in the graph database; E is the edge set; P is the property key-value pair set; src, tgt, and lbl all represent functional relationships, indicating that each edge in the graph has a start point, each edge in the graph has an end point, and each node / edge in the graph model can be set with zero or more labels to identify. 1) Node mapping S is the subject set of RDF, and O is the object set of RDF. Map S and O to V, and there are no duplicate elements in V. Denote the set of classes in the RDF file as C, and map the class c to which each element in S and O in RDF belongs to the label lbl of the corresponding element in V. 2) Relationship mapping The relationship between two classes in the ontology actually represents the possible relationships between the instances belonging to these two classes, and is mapped to the relationships in the Neo4j graph database. The classes in the ontology correspond to the labels in Neo4j, and the relationships in the ontology are mapped to the possible relationships between the nodes of these two labels. 3) Attribute mapping In the ontology model, data properties describe the characteristics of classes in the ontology model, directly corresponding to the property key-value pairs P of Neo4j nodes; Method for constructing a domain knowledge graph based on the Neo4J graph database: According to the mapping rules from ontology to Neo4j, organize the immovable cultural relic risk assessment data according to the immovable cultural relic risk assessment ontology and store it in the Neo4J graph database.

2. The method for constructing a risk assessment knowledge graph of immovable cultural relics according to claim 1, characterized in that The immovable cultural relic risk assessment ontology model has a total of eight steps, namely clarifying the domain scope, examining the possibility of reusing existing ontologies, listing important terms in the ontology, defining the hierarchical relationship between classes, defining the properties of classes and property constraints, ontology instantiation, ontology evaluation, ontology establishment, and iteration.

3. A method for constructing a knowledge graph for the risk assessment of immovable cultural relics according to claim 1, characterized in that, Text data is extracted by keyword extraction, regular matching, and manual methods, and the relevant data contained in the text data is extracted and saved in the form of an Excel table; Network data includes websites of immovable cultural relic data, and the content of these websites is extracted through web parsing technology, and the obtained content is stored in the form of an Excel table; For structured data, standardize it according to the data standardization specifications provided by experts in the field of immovable cultural relic risk assessment.

Images