Large search engines process thousands of queries per second over billions of documents, making a huge performance gap between disjunctive and conjunctive queries. An important class of optimization techniques called top-k processing is therefore used to narrow the gap. In this paper, we propose an aggressive algorithm based on the document-at-a-time (DAAT) MaxScore, aiming at further reducing the query latency of disjunctive queries. Essentially, our approach, named Aggressive MaxScore (AMaxScore), can speed up quickly by fine-tuning the initial top-k threshold, which allows a first aggressive process and then a supplementary process if not enough results are returned. Experiments with TREC GOV2 collection show that our approach reduces disjunctive query processing time by almost 15.4% on average over the state-of-the-art MaxScore baseline, while still returns the same results as the disjunctive evaluation. © 2014 Springer International Publishing Switzerland.
CITATION STYLE
Jiang, K., Song, X., & Yang, Y. (2014). Faster MaxScore document retrieval with aggressive processing. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8485 LNCS, pp. 1–4). Springer Verlag. https://doi.org/10.1007/978-3-319-08010-9_1
Mendeley helps you to discover research relevant for your work.