An efficient P2P searching and indexing strategy for multi-keyword query

Abstract

Multi-keyword search in a P2P repository system based on DHT suffers from high traffic cost due to intersection operation of distributed inverted lists. Indexing by a set of keywords (term set) is one possible solution to reduce the cost. However, as the number of terms for each set increases, the number of possible term sets increases tremendously. Index pruning partly solves the problem, with the cost of reduced recall rate. In this paper, a multi-keyword search strategy for use with the term-set index pruning technique is proposed to increase recall rate, while retaining the same index size required for the case of a two-keyword search. Experimental result supports the predicted improvement.