Topic Tracking with the PRISE Information Retrieval System
Douglas W. Oard · 2009
Information retrieval systems offer an attractive alternative to construction of a topic tracking system from scratch. The freely available PRISE vector space text retrieval system was applied to the TDT-2 topic tracking task. A simple version of the Rocchio formula was used for profile formulation and a retrieval status value threshold was used in conjunction with a temporal cutoff to make hard decisions. The results indicate that a more sophisticated approach will be needed to produce a competitive system. The paper concludes by identifying two promising directions for further work. 1. Introduction The topic tracking problem exhibits strong similarities to what has been called information filtering in the field of information retrieval [4]. In both cases, the goal is to process information objects arriving in a stream from some source based at least in part on observations of the user's reactions to previously seen objects. Two major variants of the text filtering problem exist, on...