UMass at TREC 2003: HARD and QA
MASSACHUSETTS UNIV AMHERST CENTER FOR INTELLIGENT INFORMATION RETRIEVAL
Pagination or Media Count:
The Center for Intelligent Information Retrieval CIIR at UMass Amherst participated in two tracks for TREC 2003 High Accuracy Retrieval from Documents HARD and Question Answering QA. In the HARD track, we developed document metadata to correspond to query metadata requirements implemented clarification forms based on query expansion, passage retrieval, and clustering and retrieved variable length passages deemed most likely to be relevant. This work is discussed at length in Section 1. In the QA track, we focused on retrieving passages that were likely to contain the answer to the question. In the QA track, we developed a dynamic passaging retrieval system to identify passages likely to contain answers. CIIR last participated in the QA task in TREC 9 2000. At that time we fielded the Marsha system 11. This system was based on an INQUERY document retrieval engine followed by the application of a series of heuristics rules to identify 250-byte long passages in the retrieved documents that were likely to contain the desired answers. This years passage sub-task in the QA track has allowed us to participate once again utilizing our current approach to passage retrieval with language models. We developed a dynamic passaging system that retrieved document passages based on the simplest implementation of language models cross-entropy between bag-of-word models for a question and a candidate passage.
- Information Science