A Subsequence Interleaving Model for Sequential Pattern Mining

Abstract

Recent sequential pattern mining methods have used the minimum description length (MDL) principle to define an encoding scheme which describes an algorithm for mining the most compressing patterns in a database. We present a novel subsequence interleaving model based on a probabilistic model of the sequence database, which allows us to search for the most compressing set of patterns without designing a specific encoding scheme. Our proposed algorithm is able to efficiently mine the most relevant sequential patterns and rank them using an associated measure of interestingness. The efficient inference in our model is a direct result of our use of a structural expectation-maximization framework, in which the expectation-step takes the form of a submodular optimization problem subject to a coverage constraint. We show on both synthetic and real world datasets that our model mines a set of sequential patterns with low spuriousness and redundancy, high interpretability and usefulness in real-world applications. Furthermore, we demonstrate that the quality of the patterns from our approach is comparable to, if not better than, existing state of the art sequential pattern mining algorithms.

Parameter-free probabilistic API mining across GitHub

“…Happily, this problem can be approximately solved using a simple greedy algorithm (cf. Algorithm 1 in [15]), and we find that the greedy algorithm works well in practice. The greedy algorithm repeatedly chooses an API pattern S that maximizes the improvement in log probability divided by the number of methods in S that have not yet been explained.…”

Free-Rider Episode Screening via Dual Partition Model

“…One of the defects in frequent pattern mining is that there are abundant redundant patterns in the very large number of output patterns [22]. As a result, how to effectively reduce redundancy of the output becomes an essential problem of current research [23,13,11,18,16,10,6,8]. Frequent episode mining [14] (FEM for short), as one of the sub-topics of frequent pattern mining, which aims at discovering frequently appeared ordered sets of events from a single symbol (event) sequence, is facing the similar problem as well.…”

Efficiently Summarising Event Sequences with Rich Interleaving Patterns

“…10 and 50 patterns of 5 events long 10 times each over an otherwise independent sequence, with a 10% probability of having a gap between consecutive events. To evaluate the ability of SQUISH to discover interleaved and nested patterns, we consider the Parallel database [4]. Each event in this database is generated by five independent parallel processes chosen at random.…”

“…Recently, Fowkes and Sutton proposed the ISM algorithm [4]. ISM is based on a generative probabilistic model of the sequence database, and uses EM to search for that set of patterns that is most likely to generate the database.…”