Using SAT and SQL for Pattern Mining in Relational Databases - Archive ouverte HAL Access content directly
Conference Papers Year :

Using SAT and SQL for Pattern Mining in Relational Databases

Emmanuel Coquery
Jean-Marc Petit
Lakhdar Saïs

Abstract

In this paper, we present an ongoing work bridging the gap between pattern mining, SQL and SAT for a particular class of patterns. We extend the work presented in [2] that proposes a logical query language for rule patterns satisfying Armstrong's axioms. Our contributions are the following: firstly, we allow a large part of the relational tuple calculus (SQL) to be used in the specification of queries. Secondly, we propose a boolean encoding of the query that can be used to compute answers even in the case of non Armstrongcompliant queries. Some experiments have been performed on top of Derby (embedded Java DBMS) and a modified version of MiniSat to show the feasibility of the approach.
Fichier principal
Vignette du fichier
cocomile2012_satsql.pdf (319.1 Ko) Télécharger le fichier
Origin : Files produced by the author(s)

Dates and versions

hal-00870949 , version 1 (19-01-2021)

Identifiers

  • HAL Id : hal-00870949 , version 1

Cite

Emmanuel Coquery, Jean-Marc Petit, Lakhdar Saïs. Using SAT and SQL for Pattern Mining in Relational Databases. 1st Workshop on COmbining COnstraint solving with MIning and LEarning ({CoCoMile}'12), 2012, Montpellier, France. pp.41-46. ⟨hal-00870949⟩
83 View
19 Download

Share

Gmail Facebook Twitter LinkedIn More