Distributed Data Mining by associated rules: Improvement of the Count Distribution algorithm
Today, most large systems are overwhelmed by a flood of data that is stored daily in databases distributed. It is in this context that the distributed data mining is used by offering many parallel and distributed algorithms to extract crucial information.
Among the most popular techniques, we are interested in our work at association rules technique by focusing on the distributed approach the Count Distribution (CD). We aim for our contributions to improve this algorithm by reducing the number of exchanged messages, and the number of generated candidates.
Our algorithm is based on the sequential algorithm AClose of the closed frequents itemsets approch. The experimental results showed that the proposed algorithm meets the expected objectives by presenting a performance gain greater than the CD algorithm in which the last points are important performance factors in determining the quality of an algorithm for extraction rules.
Keywords: Association rules, distributed system, the distribution count approach, closed itemsets
Download Full-Text
ABOUT THE AUTHORS
Hadj-Tayeb Karima
received the Magister degree in computer science from the University of Sciences and Technology of Oran (Algeria) in 2009. Currently, she is an assistant teaching in English department in the Es-Senia University of Oran and she is a doctorate candidate. Her research focuses on data mining and the distributed system.
Hadj-Tayeb Lilia
received the Magister degree in computer science from the University of Es-Senia of Oran (Algeria) in 2008. Currently, she is an assistant teaching in the department of computer science in the Es-Senia University of Oran and she is a doctorate candidate. Her research focuses on data mining and the Grid.
Hadj-Tayeb Karima
received the Magister degree in computer science from the University of Sciences and Technology of Oran (Algeria) in 2009. Currently, she is an assistant teaching in English department in the Es-Senia University of Oran and she is a doctorate candidate. Her research focuses on data mining and the distributed system.
Hadj-Tayeb Lilia
received the Magister degree in computer science from the University of Es-Senia of Oran (Algeria) in 2008. Currently, she is an assistant teaching in the department of computer science in the Es-Senia University of Oran and she is a doctorate candidate. Her research focuses on data mining and the Grid.