Abstract:

Association rule is a very interesting technique to discover hidden associations between attributes. These associations are very use- ful, efficient in decision-making. Fuzzy association rules have been in- troduced to treat numeric dataset and to provide more understandable representation of the knowledge. It uses fuzzy logic to transform numeric data to fuzzy data. This transformation maintains the integrity of infor- mation conveyed by the numeric attributes. Many algorithms have been presented. Unfortunately, the majority of these algorithms suffer from a huge number of extracted fuzzy association rules and a high runtime. To overcome such problems, we propose a new algorithm for fuzzy associa- tion rules mining adopting an encoding system based on prime numbers. This algorithm is able to extract generic bases of fuzzy association rules using fuzzy formal concepts analysis. To validate our algorithm, we make some experimental evaluations on some datasets.