Skip to Main content Skip to Navigation
Theses

Extraction et usages de motifs minimaux en fouille de données, contribution au domaine des hypergraphes

Abstract : Pattern discovery is a significant field of knowledge discovery in databases. This work deals with mining and using minimal generators (also called free or key patterns). First, we propose an efficient algorithm for mining &-free patterns in large databases. This is a difficult task due to the huge search space.
We presents a new approach based on pattern extension and a new pruning criterion. Second, we provide a unified view of objective interestingness measures. We design a framework capturing the main features of interestingness measures and we prove that a large set of usual measures, called SBMs behave in a similar way. We also give an algorithm to efficiently mine non-redundant rules simultaneously optimizing all the SBMs by using the free patterns. Finally, we deepen the relationship between data mining and hypergraph. We show how to exploit the key ideas of our extension-based method for efficiently computing the minimal transversals of a hypergraph which is know as a very hard problem. Experiments prove that our methods are very efficient in practice and useful for various applications.
Document type :
Theses
Complete list of metadatas

Cited literature [4 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00253794
Contributor : Hal System <>
Submitted on : Wednesday, February 13, 2008 - 10:48:07 AM
Last modification on : Monday, October 19, 2020 - 11:11:03 AM
Long-term archiving on: : Tuesday, May 18, 2010 - 12:33:54 AM

Identifiers

  • HAL Id : tel-00253794, version 1

Citation

Céline Hébert. Extraction et usages de motifs minimaux en fouille de données, contribution au domaine des hypergraphes. Autre [cs.OH]. Université de Caen, 2007. Français. ⟨tel-00253794⟩

Share

Metrics

Record views

409

Files downloads

215