Finding Related Entities by Retrieving Relations: UIUC at TREC 2009 Entity Track

摘要

Our goal in participating in the TREC 2009 Entity Trackwas to study whether relation extraction techniques can helpin improving accuracy of the entity ???nding task. Findingrelated entities is informational in nature and we wantedto explore if inducing structure on the queries helps satisfythis information need. The research outlook we took wasto study techniques that retrieve relations between two entities from a large corpus, and from those, ???nd the mostrelevant entities that participate in the given relation withanother given entity. Instead of aiming at retrieving pagesabout speci???c entities, we tried to address the problem of directly ???nding the entities from the text. Our experimentalresults show that we were able to ???nd many related entitiesusing relation-based extraction, and ranking entities basedon further evidence from the text helps to a certain extent.