| In today's explosive growth of network information,querying information through search engines has become the main way of information retrieval.However,the universality of entity renaming has led to great ambiguity in search results.How to effectively eliminate ambiguity beeomes a problem that search engines need to solve urgently,while entity disambiguation is the key technology to solve this core problem.Combined with multiple data sources,this paper proposes a named entity disambiguation algorithm based on multi-primary-attribute classified and structured semantic relationship.The algorithm does research on the disambiguation of researchers,and integrates the scattered information of multiple data sources such as China Knowledge Network and Baidu Encyclopedia.This paper extracts entities' multi-primary-attributes,successively calculates the structured and the classified semantic relationship between entities,finally obtains the similarity between entities,and use clustering algorithm to achieve entity disambiguation.The structured semantic relational algorithm mines the explicit and latent semantic relations between entities by constructing semantic relation graphs.The classified semantic relation algorithm calculates the classified semantic relationship of entities by extracting the multi-primary and non-primary attributes of the entity.Based on the algorithm,this paper finally implements a multi-source named entity disambiguation system for researchers.This system receives the user's input about the retrieval information of scientific researchers,and eliminates ambiguity with the collected scientific researcher entity data set,finally retuoms the disambiguation result.Experiments show that the named entity disambiguation algorithm based on multi-primary-attribute classification and structured semantic relationship is applied to the disambiguation of researchers,which greatly improves the accuracy of the scientific researchers information retrieval system. |
PDF Full Text Request