A Proposition for Sequence Mining Using Pattern Structures

Abstract

In this article we present a novel approach to rare sequence mining using pattern structures. Particularly, we are interested in mining closed sequences, a type of maximal sub-element which allows providing a succinct description of the patterns in a sequence database. We present and describe a sequence pattern structure model in which rare closed subsequences can be easily encoded. We also propose a discussion and characterization of the search space of closed sequences and, through the notion of sequence alignments, provide an intuitive implementation of a similarity operator for the sequence pattern structure based on directed acyclic graphs. Finally, we provide an experimental evaluation of our approach in comparison with state-of-the-art closed sequence mining algorithms showing that our approach can largely outperform them when dealing with large regions of the search space.

RCA-Seq: An original approach for enhancing the analysis of sequential data based on hierarchies of multilevel closed partially-ordered patterns

“…The pattern structure is (P, (S, ), δ), where P is the set of patients, S is a set of sequences and their subsequences, and is the set intersection. Each patient of P is described by a sequence (and its subsequences) through δ relation.This approach is deepened in [16], where object descriptions are organised into a semi-lattice of closed sets of closed subsequences, which are built based on the corresponding CPO-patterns extracted from a sequence dataset (see a comparison with our approach in Sect. 7.1).…”

“…Lines [5,7] have the complexity O(p) since C next contains p concepts pointed by Y m temporal relational attributes. Lines [8][9][10][11][12][13][14][15][16][17][18][19][20][21] are executed p times since each temporal concept of L Kt is visited only once and the complexity of these lines is O(p(q + p)). Indeed, Lines [12,14,16] are O(p) since C next has p concepts pointed by Y t temporal relational attributes; Line [10] is O(q); and the other lines are O(1).…”

“…Lattices Here, we compare the results obtained by using two distinct approaches for analysing sequential data, namely our RCA-Seq and the one introduced in [16], which relies on pattern structures. This comparison is relevant since [16] also proposes to represent the closed subsequences that coexist in one or more sequences as a directed acyclic graph of alignments, i.e. a CPO-pattern.…”

“…We use the small sequence database D S shown in Tab. 11(a), which is taken from [16]. No threshold for the support measure is employed.…”

“…Relational context family 3. comparing RCA-Seq with a pattern-structure approach [16] and with a sequential data mining approach [14]; 4. showing through an illustrative example that the structure of a hierarchy of multilevel CPO-patterns can be exploited to facilitate sequential data analysis; 5. extending RCA-Seq to integrate a user-defined taxonomy, and to extract CPO-patterns with user-specified constraints on the order relations on itemsets.…”