MF-Chord: Supporting multi-attribute multi-keyword fuzzy-matching queries

Most educational resource grids are required to support multi-attribute multi-keyword fuzzy-matching queries. But such queries are not efficiently supported in current structured P2P systems. Towards an efficient P2P system capable of processing multi-attribute multi-keyword fuzzy-matching queries with high recall ratio and load balancing, we propose a new resource indexing model which is expanded from chord and called MF-Chord. Besides one-dimensional fingerprint which includes main keywords information of each attribute of the resource, MF-Chord generates opposite-fingerprint for every resource, which is the partial reversal of the fingerprint and is used to balance the query load. Reforming the finger tables of nodes and the query-request-message format, MF-Chord dynamically generates the query-request-forwarding-tree for every query and realizes the efficient multi-attribute multi-keyword fuzzy-matching query function. Through theoretical analysis, we prove that MF-Chord has high recall ratio in limited hops. The experiment results show that the recall ratio of MF-Chord is more than 80% even when the maximum hop count is set to 7 and there are 80000 nodes in the system.