Research On Visual Analysis Method Of Knowledge Graph Based On Relationship And Structure

As Knowledge Graphs(KG)are widely used in various industries,it is crucial to improve the readability of KG and increase its acceptance by the public.In order to let the public understand KG more comprehensively,we proposed a visual analysis method of KG and verified,evaluated,and applied the method.This thesis mainly completes the following research:1.This thesis firstly conducts an in-depth analysis of the existing research around the three fields of KG representation learning technology,visual representation of knowledge graph and data visualization,and then compares their advantages and disadvantages,and finally forms a research model of knowledge graph visualization analysis.2.According to the research model of visual analysis of KG,this thesis extracts the visual analysis dimension of KG from the relationship and structure of KG itself.Secondly,this thesis defines each relationship and structure of KG and its visual analysis methods in detail.Finally,this thesis summarizes the visual analysis methodology of knowledge graph based on relational structure.3.Based on the proposed visualization analysis methodology of KG,the algorithm in the methodology is designed in detail and verified on real data.For the low-dimensional KG,this thesis proposes a data conversion algorithm for each relational structure,and uses the Beijing cultural KG data as the support to verify the algorithm theoretically.For high-dimensional KG,this thesis firstly uses the dimensionality reduction method based on the graph embedding model,and then proposes the relationship and structure definition algorithm of high-dimensional KG on the basis of dimensionality reduction,and then proposes two combined visualization layout schemes for high-dimensional KG,and finally verifies the algorithm with public datasets.4.This thesis evaluates and applies visual analysis methods for KG.In this thesis,we first use information extraction experiments to evaluate the usability of the method from the user’s perspective.Secondly,this thesis implements a system for visual analysis of KG to apply the method in feasibility.The results show that this method can effectively assist users to understand KG from multiple perspectives more comprehensively.The core of the work of this thesis revolves around the relationship and structure of KG,and strives to enrich the expression of KG through visualization technology,thereby lowering the threshold for understanding and using KG,and allowing relevant researchers to analyze KG more comprehensively.