@techreport{TD:100208,
	att_abstract={{Question-Answering is an important topic in information retrieval and natural language processing. IBM’s Watson machine [1], Microsoft’s AskMSR [2], and AT&T’s Qme! [3] are examples of QA systems which retrieve answers to questions by using machine learning techniques and by exploiting huge corpora of available resources.
Qme! is a QA system designed by Mishra and Bangalore [3], and uses millions of question/answer pairs. The provided questions are all answered and categorized by human experts. Given a user query, Qme! ranks the available questions based on their BLEU score similarity with the user query, and outputs the corresponding answers of the top 5-ranked questions. This approach works well if the new question already exists in the questions repository, and thus has a high BLEU score. However, since this approach is not based on semantic analysis of the questions, the top-ranked questions may not be semantically relevant to the user query.
The broader goal of this work is to retrieve questions that are semantically relevant to a given user query, with the aim of increasing recall. We hypothesize that questions retrieved using the expanded query will have a higher F-measure than those retrieved using the unexpanded query. Therefore, in this work, we will focus on comparing different methods for query expansion.}},
	att_authors={ph2326, sb7658, hk1971, tm330a},
	att_categories={},
	att_copyright={{}},
	att_copyright_notice={{}},
	att_donotupload={true},
	att_private={false},
	att_projects={},
	att_tags={},
	att_techdoc={true},
	att_techdoc_key={TD:100208},
	att_url={},
	author={Patrick Haffner and Srinivas Bangalore and Howard Karloff and Taniya Mishra and Sina Jafarpour},
	institution={{5th Annual Machine Learning Symposium}},
	month={October},
	title={{Learning Query Expansion in Question Answering Systems}},
	year=2010,
}