论文信息 - An n-Gram and Initial Description Based Approach for Entity Ranking Track

An n-Gram and Initial Description Based Approach for Entity Ranking Track

The most important work that takes the center stage in the Entity Ranking track of INEX is proper query formation. Both the subtasks, namely Entity Ranking and List Completion, would immensely benefit if the given query can be expanded with more relevant terms, thereby improving the efficiency of the search engine. This paper stresses on the correct identification of "Meaningful n-grams" from the given title and proper selection of the "Prominent n-grams" among them as the utmost important task that improves query formation and hence improves the efficiencies of the overall Entity Ranking tasks. We also exploit the Initial Descriptions (IDES) of the Wikipedia articles for ranking the retrieved answers based on their similarities with the given topic. List completion task is further aided by the related Wikipedia articles that boosted the score of retrieved answers.

Saswati Mukherjee | Meenakshi Sundaram Murugeshan | S. Mukherjee

[1] Grace Hui Yang,et al. Web-based List Question Answering , 2004, COLING.

[2] Ellen M. Voorhees,et al. Overview of the TREC 2002 Question Answering Track , 2003, TREC.

[3] Horacio Saggion,et al. The University of Sheffield's TREC 2003 Q&A Experiments , 2003, TREC.

[4] Elizabeth D. Liddy,et al. Question Answering: CNLP at the TREC 2002 Question Answering Track , 2002, TREC.

[5] Kentaro Torisawa,et al. Exploiting Wikipedia as External Knowledge for Named Entity Recognition , 2007, EMNLP.

[6] Ellen M. Voorhees,et al. The Tenth Text REtrieval Conference, TREC 2001 | NIST , 2002 .

[7] Eduard Hovy,et al. Knowledge-Based Question Answering , 2002 .

[8] Eduard H. Hovy,et al. Automatic Evaluation of Summaries Using N-gram Co-occurrence Statistics , 2003, NAACL.

[9] Eric Brill,et al. A Unified Framework For Automatic Evaluation Using 4-Gram Co-occurrence Statistics , 2004, ACL.

[10] Nick Craswell,et al. Overview of the TREC 2005 Enterprise Track , 2005, TREC.

[11] Philipp Koehn,et al. Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning (EMNLP-CoNLL) , 2007 .