A Probabilistic Algorithm for Multi­Relational Data Mining

José E. Ochoa LunaFabio G. Cozman

Inductive Logic Programming (ILP) algorithms learn a set of first­ order logical rules from multi­relational data, and are thus well suited to several data mining tasks. In this work, we introduce a Probabilistic ILP algorithm to extract useful regularities from multiple data tables. We adopt a probabilistic cover approach that allows us to guide the search for rules; Noisy­OR functions are employed to encode probability distributions. Preliminary tests have been conducted on relational data in the Lattes curriculum platform.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: