As stable as you are: Re-ranking search results using query-drift analysis

Abstract

This work studies the merits of using query-drift analysis for search re-ranking. A relationship between the ability to predict the quality of a result list retrieved by an arbitrary method, as manifested by its estimated query-drift, and the ability to improve that method's initial retrieval by re-ranking documents in the list based on such prediction is established. A novel document property, termed "aspectstability", is identified as the main enabler for transforming the output of an aspect-level query-drift analysis into concrete document scores for search re-ranking. Using an evaluation with various TREC corpora with common baseline retrieval methods, the potential of the proposed re-ranking approach is demonstrated.