Comparison of Two Interactive Search Refinement Techniques

Comparison of Two Interactive Search Refinement Techniques Olga Vechtomova Department of Management Sciences University of Waterloo 200 University Avenue West, Waterloo, Canada ovechtom@engmail.uwaterloo.ca Murat Karamuftuoglu Department of Computer Engineering Bilkent University 06800 Bilkent Ankara, Turkey hmk@cs.bilkent.edu.tr Abstract The paper presents two approaches to interactively refining user search formulations and their evaluation in the new High Accuracy Retrieval from Documents (HARD) track of TREC-12. One method consists of asking the user to select a number of sentences that may represent relevant documents, and then using the documents, whose sentences were selected for query expansion. The second method consists of showing to the user a list of noun phrases, extracted from the initial document set, and then expanding the query with the terms from the phrases selected by the user. 1 Introduction Query expansion following relevance feedback is a well established technique in information retrieval, which aims at improving user search performance. It combines user and system effort towards selecting and adding extra terms to the original query. The traditional model of query expansion following relevance feedback is as follows: the user reads a representation of a retrieved document, typically its full-text or abstract, and provides the system with a binary relevance judgement. After that the system extracts query expansion terms from the document, which are then added to the query either manually by the searcher – interactive query expansion, or automatically – automatic query expansion. Intuitively interactive query expansion should produce better results than automatic, however this is not consistently so (Beaulieu 1997, Koenemann and Belkin 1996, Ruthven 2003). In this paper we present two new approaches to automatic and interactive query expansion, which we developed and tested within the framework of the High Accuracy Retrieval from Documents (HARD) track of TREC (Text Retrieval Conference). 1.1 HARD track The main goal of the new HARD track in TREC-12 is to explore what techniques could be used to improve search results by using two types of information: 1. Extra-linguistic contextual information about the user and the information need, which was provided by track organisers in the form of metadata. It specifies the following: � Genre – the type of documents that the searcher is looking for. It has the following values: - Overview (general news related to the topic); - Reaction (news commentary on the topic); - I-Reaction (as above, but about non-US commentary) - Any. � Purpose of the user’s search, which has one of the following values: - Background (the searcher is interested in the background information for the topic); - Details (the searcher is interested in the details of the topic); - Answer (the searcher wants to know the answer to a specific question); - Any. � Familiarity of the user with the topic on a five- point scale. � Granularity – the amount of text the user is expecting in response to the query. It has the following values: Document, Passage, Sentence, Phrase, Any. � Related text – sample relevant text found by the users from any source, …

Content truncated to 500 words