A New Approach to Improve Association Rules for Big
Data in Cloud Environment
Djilali Dahmani, Sidi Ahmed Rahal,
and Ghalem Belalem
Department of Computer Science, University of Science and Technology,
Algeria
Abstract: The technique of association rules is very useful
in Data Mining, but it generates a huge number of rules. So, a manual post-processing
is required to target only the interesting rules. Several researchers suggest
integrating users' knowledge by using ontology and rule patterns, and then select
automatically the interesting rules after generating all possible rules.
However, nowadays the business data are extremely increasing, and many
companies have already opted for Big Data systems deployed in cloud environments,
then the process of generating association rules becomes very hard. To deal
with this issue, we propose an approach using ontology with rule patterns to
integrate users' knowledge early in the preprocessing step before searching or
generating any rule. So, only the interesting rules which respect the rule
patterns will be generated. This approach allows reducing execution time and
minimizing the cost of the post-processing especially for Big Data. To confirm
the performance results, experiments are carried out on Not Only Strutured Query
Language (NoSQL) databases which are distributed in a cloud environment.
Keywords: Big data, association rules, rule
patterns, ontology, cloud computing, NoSQL.