KeyQuery – A Front End for the Automatic Translation of Keywords into Structured Queries.pdfVIP

KeyQuery – A Front End for the Automatic Translation of Keywords into Structured Queries Martin Erwig and Jianglin He School of EECS, Oregon State University, Corvallis, OR 97331 {erwig,heji}@ Abstract: We demonstrate an approach to transform keyword queries automatically into queries that combine keywords appropriately by boolean operations, such as and and or. Our approach is based on an analysis of relationships between the keywords using a taxonomy. The transformed queries will be sent to a search engine, and the returned results will be presented to the user. We evaluate the effectiveness of our approach by comparing the precision of the results returned for the generated query with the precision of the result for the original query. Our experiments indicate that our approach can improve the precision of the results considerably. 1. Introduction The most common way to find information on the Internet is to use a search engine and provide a list of keywords describing the sought information. A problem with keyword-based search is that it often returns too many irrelevant results. The use of boolean operators can improve the preciseness of queries considerably. However, there is strong empirical evidence that end users, who are by far the largest group of users of search engines, are not able to use Boolean operators correctly. For instance, Pane and Myers found that sometimes, when people say and, they actually mean or [11] (for example, “I am interested in blue and red cars” usually expresses the interest in cars that are red or blue.) This and/or confusion happens very frequently when keywords are used that are closely related in a concept hierarchy (for example, red and blue are bot