Predicting the cumulative effect of multiple query formulations
Isak Taksa · 2005
Multiple query formulations (MQF) have been extensively explored and when used, provide considerable improvements in quality of information retrieval on the Web. However, the question of how to formulate a variety of distinctive queries and how many queries are needed in each unique case remains essentially unanswered. In this research we present a process for selecting search terms and formulating multiple short queries from the original long query. We introduce a scaled cumulative query weight (t) function which is based exclusively on the submitted long query. We demonstrate that this function can serve as a predictive variable for the effectiveness of various multiple query formulation methods and can be used algorithmically to determine the operational parameters for a multiple query formulation process.