ScholarWorks@숙명여자대학교

초록

Data stream clustering over sliding windows generates clusters as the window moves. However, iterative clustering using all data in a window is highly inefficient in terms of memory use and computational load. In this paper, we improve data stream clustering over sliding windows using sliding window aggregation and nearest neighbor search techniques. Our algorithm constructs and maintains temporal group features as a summary of the window using the sliding window aggregation technique. In order to maintain a constant size for the summary, the algorithm reduces the size of the summary by joining the nearest neighbor. We exploit locality-sensitive hashing for rapid nearest neighbor searching. In addition, we also suggest a re-clustering policy that determines whether to append a new summary to pre-existing clusters or to perform clustering on the whole summary. We conduct experiments on real-world and synthetic datasets in order to demonstrate that our algorithm can significantly improve continuous clustering on data streams with sliding windows.

키워드