查看文章

This paper describes the use of the DMINR Named entity extraction and linking system in TREC 2019. The track entered for are: News track, involves both Background Linking and Entity Ranking. Our approach to each of these tasks draws on prior work done by City, University of London at the TREC conference. In the background linking task, we treated the problems as an adhoc search task, using Named Entities (NEs) from a set of documents identified in pseudo relevance feedback, and optimizing these using a Hill-Climbing algorithm to provide a set of related articles. In the Entity Ranking task, we compared an approach using the BM25 ranking method with a probabilistic model that uses Wikipedia data as a resource to rank entities. The probabilistic model utilises an entity modeling approach to disambiguate NEs. Then, it utilises a scoring model which uses the given entities to provide a score for them based on evidence from the news articles.