Extraire et valider les relations complexes en sciences humaines : statistiques, motifs et règles d'association

Martine Cadot 1, 2
2 PAROLE - Analysis, perception and recognition of speech
INRIA Lorraine, LORIA - Laboratoire Lorrain de Recherche en Informatique et ses Applications
Abstract : This thesis is about of Data Mining in Humanistic. This branch of Artificial Intelligence is a set of methods for extracting knowledge from electronic data. Among them, the itemsets and association rules extraction is a method to build a symbolic representation of the data structure, like the classical statistical methods makes, but, unlike these ones, it can work with complex and huge data. Therefore, this computer science model, obtained by counting of cooccurrences, is not easily used by scientists : it works with dichotomics data (True/False), the interpretation of its direct results is difficult, and its validity can seem of doubt for researchers working with statistics. We propose three techniques we constructed and experimented on real data to facilitate the use of the itemsets and association rules extraction by scientists : 1) With our randomisation test based on " exchanges in cascade " in the matrix subjects x properties, one can obtain the statistically significant links between properties 2) Our fuzzification of the itemsets and association rules extraction produces fuzzy association rules close to the fuzzy rules defined by researchers of fuzzy community around Zadeh 3) With our algorithm Midova one can only extract interactions, and 4) With our meta-rules, one can clean the association rules set of its principal contradictions and redundancies
Complete list of metadatas

Cited literature [229 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00594174
Contributor : Martine Cadot <>
Submitted on : Thursday, May 19, 2011 - 9:38:54 AM
Last modification on : Monday, April 16, 2018 - 10:41:47 AM
Long-term archiving on : Saturday, August 20, 2011 - 2:21:01 AM

Identifiers

  • HAL Id : tel-00594174, version 1

Citation

Martine Cadot. Extraire et valider les relations complexes en sciences humaines : statistiques, motifs et règles d'association. Interface homme-machine [cs.HC]. Université de Franche-Comté, 2006. Français. ⟨tel-00594174⟩

Share

Metrics

Record views

959

Files downloads

2069