Apriori-Roaring-Parallel: Frequent pattern mining based on compressed bitmaps with OpenMP
Nenhuma Miniatura disponível
Data
2021-01-01
Orientador
Coorientador
Pós-graduação
Curso de graduação
Título da Revista
ISSN da Revista
Título de Volume
Editor
Tipo
Trabalho apresentado em evento
Direito de acesso
Resumo
Mining association rules is a process which consists in extracting knowledge from datasets. This is a widely used technique to analyze customer purchasing patterns, and its process is segmented in two main phases: mining frequent sets and formulating association rules. Several approaches were developed for the first phase of the mining process whose main objective was to reduce execution time. However, as all available datasets are very large (Big Data), there is a limitation regarding its application in these new sets due to excessive memory usage. We propose the Apriori-Roaring-Parallel which explores parallelism in shared memory and demands less memory usage during the mining process. In order to achieve this memory usage reduction, the Apriori-Roaring-Parallel method employs compressed bitmap structures to represent the datasets. The results obtained show that the Apriori-Roaring-Parallel method uses memory efficiently when compared to other methods.
Descrição
Palavras-chave
Idioma
Inglês
Como citar
Proceedings - IEEE Symposium on Computers and Communications, v. 2021-September.