BBN at TREC 7 : Using Hidden Markov

We present a new method for information retrieval using hidden Markov models (HMMs) and relate our experience with this system on the TREC-7 ad hoc task. We develop a general framework for incorporating multiple word generation mechanisms within the same model. We then demonstrate that an extremely simple realization of this model substantially outper-forms tf :idf ranking on both the TREC-6 and TREC-7 ad hoc retrieval tasks. We go on to present several algorithmic reenements, including a novel method for performing blind feedback in the HMM framework. Together, these methods form a state-of-the-art retrieval system that ranked among the best on the TREC-7 ad hoc retrieval task, and showed extraordinary performance in development experiments on TREC-6.