CCPM: A Scalable and Noise-Resistant Closed Contiguous Sequential Patterns Mining Algorithm

Yacine Abboud 1 Anne Boyer 1 Armelle Brun 1
1 KIWI - Knowledge Information and Web Intelligence
LORIA - AIS - Department of Complex Systems, Artificial Intelligence & Robotics
Abstract : Mining closed contiguous sequential patterns has been addressed in the literature only recently, through the CCSpan algorithm. CCSpan mines a set of patterns that contains the same information than traditional sets of closed sequential patterns, while being more compact due to the contiguity. Although CCSpan outperforms closed sequential pattern mining algorithms in the general case, it does not scale well on large datasets with long sequences. Moreover, in the context of noisy datasets, the contiguity constraint prevents from mining a relevant result set. Inspired by BIDE, that has proven to be one of the most efficient closed sequential pattern mining algorithm, we propose CCPM that mines closed contiguous sequential patterns, while being scalable. Furthermore , CCPM introduces usable wildcards that address the problem of mining noisy data. Experiments show that CCPM greatly outperforms CCSpan, especially on large datasets with long sequences. In addition, they show that the wildcards allows to efficiently tackle the problem of noisy data.
Type de document :
Communication dans un congrès
13th International Conference on Machine Learning and Data Mining MLDM 2017, Jul 2017, New York, United States. 89, pp.15, 2017, 〈10.1016/j.knosys.2015.06.014〉
Liste complète des métadonnées

Littérature citée [24 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01569008
Contributeur : Armelle Brun <>
Soumis le : mercredi 26 juillet 2017 - 10:44:31
Dernière modification le : mardi 24 avril 2018 - 13:34:39

Fichier

CCPM A Scalable and Noise-Resi...
Fichiers produits par l'(les) auteur(s)

Identifiants

Citation

Yacine Abboud, Anne Boyer, Armelle Brun. CCPM: A Scalable and Noise-Resistant Closed Contiguous Sequential Patterns Mining Algorithm. 13th International Conference on Machine Learning and Data Mining MLDM 2017, Jul 2017, New York, United States. 89, pp.15, 2017, 〈10.1016/j.knosys.2015.06.014〉. 〈hal-01569008〉

Partager

Métriques

Consultations de la notice

216

Téléchargements de fichiers

76