Microsoft Cambridge at TREC 13: Web and Hard Tracks.

Abstract

(MSRC) team this year continue to explore issues in IR from a perspective very close to that of the original Okapi team, working first at City University of London, and then at MSRC. A summary of the contributions by the team, from TRECs 1 to 7 is presented in [3]. In this work, weighting schemes for ad-hoc retrieval were developed, inspired by a probabilistic interpretation of relevance; this lead, for instance, to the successful BM25 weighting function. These weighting schemes were extended to deal with pseudo relevance feedback (blind feedback). Furthermore, the Okapi team participated in most of the early interactive tracks, and also developed iterative relevance feedback strategies for the routing task. Following up on the routing work, TRECs 7–11 submissions dealt principally with the adaptive filtering task; this work is summarised in [5]. Last year MSRC entered only