Word Recognition Errors and Relevance Feedback in Spoken Query Processing

摘要

Given the typical length of queries submitted to Information Retrieval systems, it is easy to imagine that the effects of word recognition errors (WRE) in spoken queries must be severely destructive on the system's effectiveness. The effects of word recognition errors in Spoken Document Retrieval have been well studied and well reported in recent Information Retrieval literature, but much less experimental work has been devoted to studying the effects of word recognition errors in Spoken Query Processing (SQP). The experimental work reported in this paper shows that the use of classical IR techniques for SQP is quite robust to considerably high levels of WRE, in particular for long queries. Moreover, in the case of short queries, both standard relevance feedback and pseudo relevance feedback can be effectively employed to improve the effectiveness of SQP.