Construction And Development Of Event Knowledge Graph Platform

With the rapid development of the Internet industry,people's access to knowledge has triggered a change about the way of knowledge acquisition.A knowledge graph organizes a huge amount of information in the Internet in the form of a semantic network to express knowledge,which can better describe the objective world.Traditional knowledge graphs mainly describe static knowledge such as concepts,entities and relationships.It is difficult to reflect the dynamic knowledge based on events.Focusing on the process of constructing event knowledge graphs,this thesis develops and designs an Event Knowledge Graph Platform independently by applying relevant theories and techniques.This system is mainly used to build event knowledge graphs.The main functionalities include the storage and management of event knowledge,the storage and visualization application of graph data,the automatic construction of event knowledge graphs.The system is designed based on the Browser/Server structure.According to the modularity principle,the system is divided into News Collection Module,Graph Building Module,Algorithm Interaction Module,Data Persistence Module,Data Management Module,WEB Interaction Module.In this thesis,we focus on the following element tasks around the event knowledge graph platform:1.Combined with the theory related to event knowledge graph,this thesis proposes a representation method of event knowledge and realizes the transformation of event knowledge into graph data.On this basis,a storage method for structured event knowledge and semi-structured graphic data has been designed and implemented.2.By researching the construction methods of event knowledge graph,this thesis designes and implementes the methods of event relationship identification based on pattern matching,event knowledge extraction based on syntactic dependency analysis,event fusion based on event similarity calculation.Two complete event knowledge graph construction schemes have been designed and implemented for different application scenarios.3.After analyzing the system requirements,the system has been implemented using a separate front and back-end development approach.The front-end interface is designed using the Vue framework.The server side is developed using Spring boot.The database technologies include Mysql,Neo4 j and Redis.The algorithm module is implemented in Python.The system builds a task message queue based on the Redis database to decouple and interact with the algorithm module and the system server.Finally,a series of system testing cases have been conducted.The test results show that the system can meet the expected functions.In the process of studying the construction of event knowledge graph,this thesis discovers and realizes some system extension functions,such as news clustering and news recommendation.