Perspectives on Information Retrieval and Speech

Abstract

Several years of research have suggested that the accuracy of spoken document retrieval systems is not adversely affected by speech recognition errors. Even with error rates of around 40%, the effectiveness of an IR system falls less than 10%. The paper hypothesizes that this robust behavior is the result of repetition of important words in the text—meaning that losing one or two occurrences is not crippling- and the result of additional related words providing a greater context- meaning that those words will match even if the seemingly critical word is misrecognized. This hypothesis is supported by examples from TREC’s SDR track, the TDT evaluation, and some work showing the impact of recognition errors on spoken queries.

The field of Information Retrieval naturally includes myriad other research issues, ranging from formal modeling of the problem to engineering systems that work across languages, from document clustering to multi-document summarization, and from classification to question answering. In this paper I will focus on search engine technology, though many of the ideas and directions apply equally well to other IR research problems.