OzU-NLP at TREC NEWS 2019: Entity ranking
Type :
Conference paper
Publication Status :
Published
Access :
restrictedAccess
Abstract
This paper presents our work and submission for TREC 2019 News Track: Entity Ranking Task. Our approach utilizes Doc2Vec's ability to represent documents as fixed sized numerical vectors. Applied on news articles and wiki-pages of the entities, Doc2Vec provides us with vector representations for these two that we can utilize to perform ranking on entities. We also investigate whether background linked articles can be useful for entity ranking task.
Source :
28th Text REtrieval Conference, TREC 2019 - Proceedings
Date :
2019
Publisher :
National Institute of Standards and Technology (NIST)
Collections
Share this page