Consistent Top-k Queries over Time

Mong-Li Lee, Wynne Hsu, Ling Li, Wee Hyong Tok

Abstract: Top-k queries have been well-studied in snapshot databases and data streams. We observe that decision-makers are often interested in a set of objects that exhibit a certain degree of consistent behavior over time. We introduce a new class of queries called consistent top-k to retrieve k objects that are always amongst the top at every time point over a specified time interval. Applying top-k methods at each time point leads to large intermediate results and wasted computations. We design two methods, rank-based and bitmap, to address these shortcomings. Experiment results indicate that the proposed methods are efficient and scalable, and consistent top-k queries are practical in real world applications.