Research And Application Of Methods For Constructing Knowledge Graphs

The characteristic of the knowledge graph is that it has a specific relational structure.It can easily describe the ontology knowledge in real life,the instances and the relationships between them.The graph structure associates information on the Internet into a form for people to learn knowledge.The knowledge graph is the key to driving the smart industry.At present,the application scenarios and methods of large-scale knowledge graphs are still relatively limited.There are very few knowledge graphs in the industry domain.Moreover,the understanding and attention angles of the knowledge graphs in various industries are different.Therefore,the established knowledge graphs are also difficult to integrate,which brings difficulties to their applications.In order to propose a method of constructing an industry knowledge graph,this thesis constructs a knowledge graph in the seismic domain,and develops a knowledge graph visualization system based on the Neo4 j graph database,and then predicts the relationship on the constructed knowledge graph.In order to realize the construction,storage,and visualization of knowledge maps in the seismic field,the following research works have been done in this subject:(1)Construct a knowledge map in the seismic field.Describes the detailed steps of constructing a knowledge map in the seismic field.First,construct the ontology.Some ontology concepts are universal.You need to add some relations and concepts according to the application scenario,and then use the data provided by professional institutions to link to the ontology network.(2)Predict the relationship based on the knowledge map in the seismic domain,and further improve and supplement the established knowledge map.In this regard,this article introduces a path sorting algorithm suitable for graph structures.By taking the associated paths between entities as features,the predicted relationships are scored and ranked,and finally the relationships with higher scores are selected as the prediction results.(3)Based on the application of the earthquake knowledge map,the earthquake building damage prediction is carried out on the basis of the established knowledge map.This chapter firstly introduces the steps of traditional earthquake damage prediction for buildings,and then proposes methods based on knowledge maps and machine learning to address its shortcomings.This method avoids that the traditional method depending on the experience level of industry experts and consuming a lot of manpower.(4)Design and implementation of the knowledge graph visualization system.This chapter designs and implements a software system to quickly build a knowledge map prototype,gives the overall module design and displays the visualization results of the constructed seismic knowledge map,which proves the feasibility and practicability of the entire prototype system.