Construction And Implementation Of Knowledge Graph Oriented To Specific Domain

As a popular technology that has emerged in recent years,knowledge graph is considered to be the foundation of cognitive intelligence.It describes concepts,entities and relationships of the real world in a structured way,which provides management and organization of massive Internet information in an effective mean.In general,knowledge graphs can be divided into open knowledge graphs and domain knowledge graphs.Open knowledge graphs include knowledge in various domains,which pay more attention to the breadth of knowledge,and domain knowledge graphs contain more domain expertise.At present,more and more knowledge graphs are used in specific domains for intelligent search,semantic question answering,etc.and it proved to be doing well.With the improvement of material and cultural level,people spend more and more in the movie market.At present,the main channel for the audience to learn about the movie is the professional website of movie,but there are shortcomings in which the search and display methods are not user-friendly.For this reason,this thesis applies the knowledge graph to the domain of movie and builds a visual movie knowledge graph system for movie query.This thesis designs a crawler programs based on the Scrapy crawler framework to capture the original data of movies from the movie website.After cleaning and sorting the data,the ontology is used to describe the schema of the knowledge graph.According to the schema of knowledge graph,this thesis builds two knowledge storage systems,one is based on RDF database and another is based on graph database.After comparing the two,RDF database is finally selected to store knowledge graph data.Then this thesis uses conditional random field model to realize movie named entity recognition,and based on the built RDF knowledge storage system,we use Distant Supervision to propose a movie-person relationship extraction method.Knowledge is extracted from unstructured text and added to the knowledge graph by means of movie entity recognition and relationship extraction.Finally,based on the movie knowledge graph constructed in the above process,this thesis implements the visual movie knowledge graph query system,describes the requirements analysis,design and implementation details of the system,then completes the testing and verification of the system.