Towards Robust Methods for Spoken Document Retrieval

In this paper, we investigate a number of robust indexing and re-trieval methods in an effort to improve spoken document retrieval performance in the presence of speech recognition errors. In par-ticular, we examine expanding the original query representation to include confusible terms; developing a new document-query retrieval measure based on approximate matching that is less sen-sitive to recognition errors; expanding the document representa-tion to include multiple recognition hypotheses; modifying the original query using automatic relevance feedback to include new terms found in the top ranked documents; and combining infor-mation from multiple subword unit representations. We study the different methods individually and then explore the effects of combining them. Experiments on radio broadcast news data show that using a combination of these methods can improve retrieval performance by over 20%. 1