A Semantic Approach for Entity Linking by Diverse Knowledge Integration incorporating Role-Based Chunking
Gerard Deepak, Naresh Kumar Devulapally, A. Santhanavijayan
Abstract
Web-data has seen an exponential rise in the past few years. With the increase in the data on the web, the process of associating entities with required knowledge becomes extremely difficult. Linking entities not only becomes a tedious task but also requires the right association of knowledge with the right techniques. With the development of the Semantic Web in recent times, semantic strategies are required to represent, reason and link entities. In this paper, an entity linking approach that rightly associates personalities has been proposed. The proposed algorithm encompasses role-based chunking along with a fragmented parse tree generation. The proposed strategy performs Entity Linking by JSON fragmented parse tree generation and recommends the entities based on the semantic score generated by computing the concept similarity. The knowledge is supplied by a role-based Ontology modeled for various famous personalities. An accuracy of 89.77% is achieved for role-based entity linking which is much better and reliable than the existing strategies, especially when a large number of trials were conducted for the Indian Context.