2012 | OriginalPaper | Buchkapitel
General Algorithms for Mining Closed Flexible Patterns under Various Equivalence Relations
verfasst von : Tomohiro I, Yuki Enokuma, Hideo Bannai, Masayuki Takeda
Erschienen in: Machine Learning and Knowledge Discovery in Databases
Verlag: Springer Berlin Heidelberg
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
We address the closed pattern discovery problem in sequential databases for the class of
flexible
patterns. We propose two techniques of coarsening existing equivalence relations on the set of patterns to obtain new equivalence relations. Our new algorithm
GenCloFlex
is a generalization of
MaxFlex
proposed by Arimura and Uno (2007) that was designed for a particular equivalence relation.
GenCloFlex
can cope with existing, as well as new equivalence relations, and we investigate the computational complexities of the algorithm for respective equivalence relations. Then, we present an improved algorithm
GenCloFlex+
based on new pruning techniques, which improve the delay time per output for some of the equivalence relations. By computational experiments on synthetic data, we show that most of the redundancies in the mined patterns are removed using the proposed equivalence relations.