An Online Static Index Pruning Algorithm

摘要

Static index pruning can significantly reduce index size and query processing time. An online static index pruning algorithm is presented, which is a term-centric method and adopts BM25 weighting as the pruning measure. The algorithm scans through documents set with one pass and directly builds pruned index, and therefore avoids the construction of original index. The experiments based on TREC data set show that the online static index pruning algorithm requires less time to build pruned index, and the pruning effectiveness outperforms the baseline method.