Research And Implementation Of Knowledge Map Construction For Automobile Domain

Since the 21 st century,people have gradually realized the importance of information and data.But now the data on the network is blowout-like growth,how to quickly search out the data of interest from the massive data is an urgent problem to be solved.The rise of knowledge atlas can help us solve this problem.Knowledge atlas can mine the semantic relationship between entities and help us better organize data.With the improvement of people's living standard,automobile has become an indispensable part of people's life,and people's demand for relevant information in the automobile field is also growing.Knowledge atlas for automotive field can facilitate people to query information,configuration,comments and other information in automotive field.This paper first introduces the research background and significance of knowledge atlas in automobile field,then expounds the current situation of knowledge atlas construction technology,and analyzes the existing problems.At present,most of the existing triple extraction algorithms extract triples directly from web text,and the quality of the triples is not very high.Substituting pronouns for the objects mentioned above is a common way of expression in Chinese.Previous algorithms often create pronouns such as "he" and "they" in triples,which greatly reduces the quality of triples.According to the characteristics of data and information in automotive field,this paper proposes a set of complete algorithm for building knowledge map in automotive field,which can solve this problem well.Firstly,Scrapy crawler framework is used to crawl the relevant data information of Sina Automobile Network,Netease Automobile Network and Pacific Automobile Network.Secondly,decision tree algorithm is used to anaphorize the crawled text,and then dependency analysis method is used to extract triples.At this time,the quality of triples obtained is relatively high.In this paper,the graph database Neo4 j is used to store triple data.Neo4 j stores triple data in the network rather than in tables,which can improve the efficiency of search.This paper uses PHP to interact with Neo4 j.The back-end framework uses Thinkphp framework based on MVC three-tier architecture.The front-end uses Echarts to do data visualization.In the visualization module of knowledge map,the functions of importing and exporting triple data and querying and searching triple data are realized,and the search results are displayed in the way of relational graph.After a lot of experiments and analysis,the proposed knowledge map construction algorithm can basically eliminate the interference caused by eliminating pronouns,thus greatly improving the accuracy of triple and the quality of knowledge map.Data videophone module is also more humane,and can see the search results more intuitively.