Efficient Fuzzy Top-k Query Processing over Uncertain Objects

Abstract

Recently, many application domains, such as sensor network monitoring and Location-Based Service, raise the issue of uncertain data management. Uncertain objects, a kind of uncertain data, have some uncertain attributes whose values are ranges instead of points. In this paper, we study a new kind of top-k queries, Probabilistic Fuzzy Top-k queries (PF-Topk queries) which can return k results from uncertain objects for fuzzy query conditions. We formally define the problem of PF-Topk query and present a framework for answering this kind of queries. We propose an exact algorithm, Envelope Planes of Membership Function (EPMF) algorithm based on the upper and lower bounding functions, which answers fuzzy top-k queries over uncertain objects in high-dimensional query space efficiently. We also propose an approximate algorithm which improves efficiency while ensuring high precision by setting a proper value of parameter. To reduce the search space, a pruning method is proposed to safely prune some objects before querying. The effectiveness and efficiency of our algorithms is demonstrated by the theoretical analysis and experiments with synthetic and real datasets.

The research was partially supported by NSF of China under Grants Nos. 60873009, 60933001 and 60773220.