Mining fuzzy association rules in data streams

Many algorithms have been proposed for mining fuzzy association rules in static datasets with quantitative attributes. However, there is few study on mining fuzzy association rules in data streams. This paper presents an algorithm FFI_Stream for fuzzy association rules mining in data streams. Efficient techniques are presented to find fuzzy association rules in data streams using time based sliding window model. The clustering technique is used to find the fuzzy sets. Membership Function Bias measure (MFB_measure) is introduced to evaluate the membership function in each sliding window. Selectively Updating Mechanism (SUM) and Projected Summaries (PS) are proposed to update the fuzzy sets dynamically. Then, experiments are carried out on both synthetic and real life datasets. The results show that the algorithm is effective and efficient.