Preference-aware sequence matching for location-based services

Sequantial data are important in many real world location based services. In this paper, we study the problem of sequence matching. Specifically, we want to identify the sequences most similar to a given sequence, under three most commonly used preferece-aware similarity measures, i.e., Fagin's intersection metric, Kendall's tau, and Spearman's footrule. We first analyze the properties of these three preference-aware similarity measures, revealing the connection between them and set intersection. Then, we build an index structure, which is essentially a doubly linked list, to facilitate efficient sequence matching. Lower- and upper-bounds are derived to achieve support prefix-based filtering. Experiments on various datasets show that our proposed method outperforms the baselines by a large margin.