Efficient Longest Streak Discovery in Multidimensional Sequence Data

This paper studies the problem of discovering longest streak in multidimensional sequence dataset. Given a multidimensional sequence dataset, the contextual longest streak is the longest consecutive tuples in a context subspace which match with a specific measure constraint. It has various applications in social network analysis, computational journalism, etc. The challenges of the longest streak discovery problem are (i) huge search space, and (ii) non-monotonicity property of streak lengths. In this paper, we propose a novel computation framework with a suite of optimization techniques for it. Our solutions outperform the baseline solution by two orders of magnitude in both real and synthetic datasets. In addition, we validate the effectiveness of our proposal by a real-world case study.