Research And Application Of Entity Aligning Method For Chinese-Korean Bilingual Knowledge Graph

In the wave of globalization,how to use artificial intelligence technology to better solve the communication and communication barriers between users of different languages has become one of the important research topics in the field of natural language processing.In addition to being used to optimize the search performance of search engines,knowledge graphs are now widely used in various professional fields such as biology,medical treatment,and finance.Researchers have benefited from the development of multilingual online encyclopedias based on Wikipedia,and they have compiled a large number of structured multilingual aligned corpora.At the same time,the embedded knowledge graph representation learning technology represented by Trans E allows researchers to design a Model to achieve automatic alignment between different language entities becomes possible.However,the current cross-language alignment methods have some shortcomings: Firstly,the alignment method based on embedding representation learning has high requirements on the scale of the data set,and the scale of the data set directly affects the accuracy of the alignment model.Secondly,if the method of combining graph convolutional neural network and representation learning is used to design the entity alignment model,the problem of under-fitting is likely to occur when facing heterogeneous knowledge graphs of different languages.On the basis of solving the above problems,this paper proposes a cross-language entity alignment model for the Chinese and Korean knowledge graph corpus.The specific research process is as follows:Firstly,more than 80,000 high-quality Chinese-Korean bilingual aligned structured data sets were collected and sorted from the Internet using crawler technology.The study of Chinese and Korean languages is an important research topic of natural language processing and linguistics.This data set can fill the gaps in the bilingual direction of the multilingual knowledge graph in the field of Chinese and Korean,and provide important basic data for the development of upstream and downstream research on the relevant knowledge graph.Secondly,an entity alignment model combining graph attention network and Trans H is proposed and implemented.The model effectively alleviates the negative impact of the small data set and the heterogeneous knowledge map on the alignment model,and improves the accuracy of the alignment model.When Korean is aligned with Chinese,Hits@1 is 48.76%,Hits@5 is 79.51%,and Hits@10 is 90.99%;when Chinese is aligned with Korean,Hits@1 is 48.55%,Hits@5 is 78.77%,and Hits@10 is90.96%.Finally,a cross-language knowledge query module was designed and implemented.This module uses the alignment model to establish relationships between bilingual entities with the same semantics,constructs a bilingual alignment knowledge graph,and uses Popoto.js and Html for front-end design,Flask framework for back-end data interaction,and realizes cross-language knowledge query.