Author : Miss Ankita Uday Manekar and Prof.Amar Chandgude

Lot of real life applications generates bulk data in streams. The analysis of real time data is required in variety of domains. The applications like market basket analysis requires the analysis of utility based co-occurrence patterns as well as frequency based co-occurrence patterns. In literature these techniques are studied on independently on data stream. The proposed work focuses on the mining of frequency based and utility based cooccurrence pattern extraction from multiple data streams. The system extracts top k patterns from multiple data streams. A sliding window protocol updates the top k co-occurrence patterns. For cooccurrence pattern extraction cp-graph, pattern utility table and inverted file structure is used. The system performance is tested on the basis of window size , processing time and memory required.

Keywords: Co-occurrence patterns, utility patterns, top k itemset, streaming data, static data, multiple stream,cp-graph