Flexible Pseudo-Relevance Feedback via Direct Mapping and Categorization of Search Requests

This paper explores various strategies for enhancing the reliability of pseudo-relevance feedback using TREC and NTCIR test collections. For each test request, the number of pseudo-relevanct documents ( ) or the number of expansion terms ( ) is determined based on a similar training request (i.e. via direct mapping) or a group of similar training requests (i.e. via categorization). The results suggest that the offer weight may be useful to some extent for determining and that categorization may be more reliable than direct mapping. These flexble pseudo-relevance feedback approaches probably deserve further investigation with larger sets of search requests.