Chinese Entity Linking Method Based On Context And Multi-Feature Fusion Graph Model

With the development of network information and the increasing needs of semantic search,knowledge base population becomes a hotspot in the field of natural language processing research.Entity linking is the key technology of knowledge base population and is the task of linking name mention in a document with their referent entities in a knowledge base,plays an important role in the theoretical research and practical application.At present,most entity linking technologies deal with the language of English,research on Chinese is still in the initial stage.There are two main reasons for this situation.(1)Lacking a unified and authoritative Chinese knowledge base and corpus;(2)Because of Chinese has rich semantic,syntax is flexible,and word segmentation is hard,Chinese entity linking technology still remains in the expression level.To solve the above problems,this paper combines with the current English entity linking technology and the research status on Chinese,proposes a method based on context and multi-feature fusion graph model:(1)In view of the first problem,this paper chooses the Chinese Wikipedia as the knowledge base of this entity linking task,extracts and constructs the Chinese corpus and experimental data sets from the official evaluation data of NIST TAC KBP;(2)In order to improve the accuracy of Chinese word segmentation and add the semantic information effectively,we extract varieties of the features to measure the semantic similarities,and model a graph which represents the relationship between the name mention and the candidate entities with these similarities from the context of name mention and the information of the candidate entities in Wikipedia,and then,use the consistency feature of graph model to rank the candidate entities and implement entity linking.In order to verify the performance of the method,using the method to reproduce the latest Chinese entity linking.The experimental results show that this method proposed in this paper can improve the accuracy and efficiency of entity linking effectively,and it obtains a better overall effect.