Skip to Main content Skip to Navigation
Theses

Une méthode de classification non-supervisée pour l'apprentissage de règles et la recherche d'information

Abstract : Data clustering is a major, but a hard, task in the unsupervised learning domain. This process is used in various context such as Knowledge Discovery, representation or description simplification of a data set.

In this study, we present the clustering algorithm PoBOC which organizes a dataset into overlapping classes which naturally match with real concepts of data. This clustering method is used in two very different applications.

- In the supervised learning field, the induction of a set of propositional and first-order rules is performed by first organizing each class into sub-classes.
- In the Information Retrieval field, the ambiguities from natural langage naturally induce overlaps between thematic.

On these two research domains, the organization of a dataset into overlapping clusters is validated with suitable experimental studies.
Document type :
Theses
Complete list of metadatas

Cited literature [64 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00084828
Contributor : Guillaume Cleuziou <>
Submitted on : Monday, July 10, 2006 - 4:48:16 PM
Last modification on : Friday, October 23, 2020 - 4:37:12 PM
Long-term archiving on: : Monday, April 5, 2010 - 11:58:30 PM

Identifiers

  • HAL Id : tel-00084828, version 1

Citation

Guillaume Cleuziou. Une méthode de classification non-supervisée pour l'apprentissage de règles et la recherche d'information. Autre [cs.OH]. Université d'Orléans, 2004. Français. ⟨tel-00084828⟩

Share

Metrics

Record views

1629

Files downloads

31035