Abstract
Many of the research tasks in Topic Detection and Tracking have counterparts in Information Retrieval research. However, the two research communities evaluate their tasks differently, which makes it very difficult to determine the extent to which they can help each other. In this study, we compare the performance of TDT tracking task to the IR filtering task, and show that they have nearly identical effectiveness. We also show a method for using tracking to predict error rates for the TDT First Story Detection (FSD) task. We then show that FSD performance is what tracking predicts. More importantly, we show that with current approaches, FSD performance has probably reached the limits of effectiveness.
Users
Please
log in to take part in the discussion (add own reviews or comments).