Mining frequent patterns from dynamic data streams with data load management

► We propose a count-approximation method for data-stream frequent-itemset mining. ► One mechanism to accelerate the processing rate is devised for overload handling. ► Another mechanism to shed the exceeding data is also designed for overload handling. ► Experimental results verify the performance of the overload-handling mechanisms.