Query Rewriting for Rule Mining in Databases

Abstract : Promoting declarative approaches in data mining is a long standing theme. This paper goes into this direction by proposing a well- founded logical query language, SafeRL, allowing the expression of a wide variety of “rules” to be discovered against the data. SafeRL ex- tends and generalizes functional dependencies in databases to new and unexpected rules easily expressed with a SQL-like language. In this set- ting, every rule mining problem turns out to be seen as a query process- ing problem. We provide a query rewriting technique and a construc- tive proof of the main query equivalence theorem, leading to an efficient query processing technique. Based on a concrete SQL-like grammar for SafeRL, we show how a tight integration can be performed on top of any DBMS. The approach has been implemented and experimented on sensor network data. This contribution is an attempt to bridge the gap between pattern mining and databases and facilitates the use of data mining techniques by SQL-aware analysts.
Type de document :
Communication dans un congrès
Bruno Crémilleux, Luc De Raedt, Paolo Frasconi, Tias Guns. Languages for Data Mining and Machine Learning (LML) Workshop@ECML/PKDD 2013, Sep 2013, Prague, Czech Republic. pp.1-16, 2013
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-01339257
Contributeur : Équipe Gestionnaire Des Publications Si Liris <>
Soumis le : mercredi 29 juin 2016 - 15:50:32
Dernière modification le : vendredi 10 novembre 2017 - 01:19:30

Identifiants

  • HAL Id : hal-01339257, version 1

Citation

Brice Chardin, Emmanuel Coquery, Benjamin Gouriou, Marie Pailloux, Jean-Marc Petit. Query Rewriting for Rule Mining in Databases. Bruno Crémilleux, Luc De Raedt, Paolo Frasconi, Tias Guns. Languages for Data Mining and Machine Learning (LML) Workshop@ECML/PKDD 2013, Sep 2013, Prague, Czech Republic. pp.1-16, 2013. 〈hal-01339257〉

Partager

Métriques

Consultations de la notice

149