Dynamic cutoff prediction in multi-stage retrieval systems

Modern multi-stage retrieval systems are comprised of a candidate generation stage followed by one or more reranking stages. In such an architecture, the quality of the final ranked list may not be sensitive to the quality of the initial candidate pool, especially in terms of early precision. This provides several opportunities to increase retrieval efficiency without significantly sacrificing effectiveness. In this paper, we explore a new approach to dynamically predicting the size of an initial result set in the candidate generation stage, which can directly affect the overall efficiency and effectiveness of the entire system. Previous work exploring this tradeoff has focused on global parameter settings that apply to all queries, even though optimal settings vary across queries. In contrast, we propose a technique that makes a parameter prediction to maximize efficiency within an effectiveness envelope on a per query basis, using only static pre-retrieval features. Experimental results show that substantial efficiency gains are achievable. In addition, our framework provides a versatile tool that can be used to estimate the effectiveness-efficiency tradeoffs that are possible before selecting and tuning algorithms to make machine-learned predictions.