Du local au global : un nouveau défi pour l'analyse statistique implicative

Abstract : Statistical implicative analysis, as well as the vast majority of current approaches towards rule mining, are faced with a certain number of difficulties such as the issue of over abundant and redundant rules. While a large number of papers in pattern mining have addressed these issues, most have concentrated on finding solutions rather than identifying causes. We present here one of the main causes of these issues: the local nature of the various approaches in rule mining, which can qualify both the models used for the measure of rule interestingness and the mining process; as well as its counterpart: choosing a global approach. We suggest a certain number of options allowing to make this choice with the objective of opening up to new perspectives and encouraging the transition from a vast majority of current local approaches towards more global approaches.
Complete list of metadatas

Cited literature [38 references]  Display  Hide  Download

https://hal.archives-ouvertes.fr/hal-01616110
Contributor : Bibliothèque Télécom Bretagne <>
Submitted on : Friday, October 13, 2017 - 10:34:54 AM
Last modification on : Tuesday, February 26, 2019 - 11:36:02 AM

File

delacroix_etal_ASI_2017.pdf
Publisher files allowed on an open archive

Identifiers

  • HAL Id : hal-01616110, version 1

Citation

Thomas Delacroix, Philippe Lenca, Stéphane Lallich. Du local au global : un nouveau défi pour l'analyse statistique implicative. ASI 2017 : 9e Colloque International d'Analyse Statistique Implicative, Oct 2017, Belfort, France. pp.103 - 116. ⟨hal-01616110⟩

Share

Metrics

Record views

240

Files downloads

50