KTV-Tree: Interactive Top-K Aggregation on Dynamic Large Dataset in the Cloud

This paper studies the problem of supporting interactive top-k aggregation query over dynamic data in the cloud. We propose TV-TREE, a top-K Threshold-based materialized View TREE, which achieves the fast processing of top-k aggregation queries by efficiently materialized views. A segment tree based structure is adopted to organize the views in a hierarchical manner. A suite of protocols are proposed for incrementally maintaining the views. Experiments are performed for evaluating the effectiveness of our solutions, in terms of query accuracy, costs and maintenance overhead.