Performing efficient keyword search by keyword grouping in DHT peer-to-peer network

The primary challenge in developing a peer-to-peer file sharing system is implementing an efficient keyword search mechanism. Current keyword search system for structured P2P systems relies on the intersection of distributed inverted index. However, when executing multiple-attribute queries, the bandwidth overhead is unacceptable. In order to reduce query overhead, indexing can be done by a set of keywords which is adopted in KSS system. However, KSS index is considerably larger than standard inverted index and the insert and storage overhead are dramatically increased. In this paper, by adopting term ranking approach such as TFIDF and exploiting the relationship information between query keywords, the indexing is done by individual keyword while search can be done in a set of related keywords like KSS. Experiments results clearly demonstrated that the improved keyword search system can match standard inverted index in insert overhead and storage overhead, while can compete with KSS index in query overhead.