A Unified Framework for Frequent Sequence Mining with Subsequence Constraints

Abstract

Frequent sequence mining methods often make use of constraints to control which subsequences should be mined. A variety of such subsequence constraints has been studied in the literature, including length, gap, span, regular-expression, and hierarchy constraints. In this article, we show that many subsequence constraints-including and beyond those considered in the literature-can be unified in a single framework. A unified treatment allows researchers to study jointly many types of subsequence constraints (instead of each one individually) and helps to improve usability of pattern mining systems for practitioners. In more detail, we propose a set of simple and intuitive "pattern expressions" to describe subsequence constraints and explore algorithms for efficiently mining frequent subsequences under such general constraints. Our algorithms translate pattern expressions to succinct finite-state transducers, which we use as computational model, and simulate these transducers in a way suitable for frequent sequence mining. Our experimental study on real-world datasets indicates that our algorithms-although more general-are efficient and, when used for sequence mining with prior constraints studied in literature, competitive to (and in some cases superior to) state-of-the-art specialized methods.

Scalable Frequent Sequence Mining with Flexible Subsequence Constraints

“…Subsequence constraints. We follow [5], [7] and express subsequence constraints using subsequence predicates of form π : Σ * × Σ * → { 0, 1 }. We say that S is a π-subsequence of T , denoted S π T , if S T and π(S, T ) = 1.…”

“…2, a subsequence predicate π ex may specify that we are interested in only the subsequences that begin with A or one of its descendants and end with b. Adopting the the language of [5], [7] (see below), we can express this constraint using pattern expression Pattern expressions are based on regular expressions, but additionally include capture groups (in parentheses), hierarchies (by omitting = ), and generalizations (using ↑ and ↑ = ). Intuitively, pattern expressions work like regular expressions: when they match, they output what is captured and may generalize along the hierarchy (optionally via ↑ , always via ↑ = ).…”

“…One approach to improve flexibility is the use of subsequence constraints, which specify conditions under which a subsequence is potentially interesting to the particular application. Ordered by increasing flexibility, common types of subsequence constraints include length constraints [28], [34], gap and duration constraints [14], [28], [34], hierarchy constraints [28], "output filter" regular expression constraints [2], [3], [13], [31], and regular expression constraints with capture groups and hierarchies [5], [7]. The latter type subsumes the remaining ones, and we subsequently refer to it as flexible constraints.…”

“…PREP?) ENTITY [12] in the pattern language of DESQ [5], [7]-but they cannot handle very large datasets. Other examples involving flexible constraints include the construction of the well-known Google n-gram corpus [30] and mining of protein sequences that exhibit a given motif [31].…”

TSARM-UDP: An Efficient Time Series Association Rules Mining Algorithm Based on Up-to-Date Patterns

“…Kaustubh Beedkar et al proposed a scalable, distributed sequence mining algorithm dealing with large amounts of data. The authors built a distributed framework for frequent sequence mining [ 24 ]. The description of temporal trends for the clinical domain of hemodialysis was proposed in [ 25 ], which considered specific temporal features with respect to the chosen time granularity.…”