An optimized interaction strategy for bayesian relevance feedback

A new algorithm and systematic evaluation is presented for searching a database via relevance feedback. It repre-sents a new image display strategy for the PicHunter system [2, 1]. The algorithm takes feedback in the form of relative judgments (“item A is more relevant than item B”) as opposed to the stronger assumption of categorical relevance judgments (“item A is relevant but item B is not”). It also exploits a learned probabilistic model of hu-man behavior to make better use of the feedback it obtains. The algorithm can be viewed as an extension of indexing schemes like the k-d tree to a stochastic setting, hence the name “stochastic-comparison search. ” In simulations, the amount of feedback required for the new algorithm scales like log2 |D|, where |D | is the size of the database, while a simple query-by-example approach scales like |D|a, where a < 1 depends on the structure of the database. This theo-retical advantage is reflected by experiments with real users on a database of 1500 stock photographs.