A pruning algorithm for optimal diversified search

Fei Chen, Yiqun Liu, Jian Li, Min Zhang, Shaoping Ma

Abstract: Given a number of possible sub-intents (also called subtopics) for a certain query and their corresponding search results, diversified search aims to return a single result list that could satisfy as many users' intents as possible. Previous studies have demonstrated that finding the optimal solution for diversified search is NP-hard. Therefore, several algorithms have been proposed to obtain a local optimal ranking with greedy approximations. In this paper, a pruned exhaustive search algorithm is proposed to decrease the complexity of the optimal search for the diversified search problem. Experimental results indicate that the proposed algorithm can decrease the computation complexity of exhaustive search without any performance loss.