From "More Like This" to "Better Than This"

Abstract

In this paper we address a novel retrieval problem we term the "Better Than This" problem. For a given pair of a user query to be answered by some search engine and a single example answer provided by the user that may or may not be a correct answer to the query, we determine whether or not there exists some better answer within the search engine. The approach we take is to test whether the user's provided answer can be used for relevance feedback in order to improve the ability of the search engine to better answer the user's query. If this is indeed the case, then we determine that the original answer provided by the user is good enough and there is no need to consider a better alternative. Otherwise, we decide that the best alternative that the search engine can provide should be considered as a better answer. Using a simulation based evaluation, we demonstrate that, our approach provides a better decision making solution to this problem, compared to several other alternatives.