A novel fuzzy association rule with Ontology concept for efficient Mining in ubiquitous real time data Applications

Abstract

Ubiquitous data mining is involved in processing a large number of datasets for information gaining. However, processing of such vast range of dataset is considered as a challenging task due to increased computational time and complexity. Existing techniques also fail to reduce computational time for ubiquitous dataset. At present, association rules exhibit significant performance for ubiquitous data mining. But processing of numerous rules and identification of ranges is difficult for processing with improved complexity. To resolve these challenges with reduced computational time, this research aimed to generate association rules for ubiquitous data mining using fuzzy. The entire research is organized into three phases. In the first phase, this research intended to develop FARs (Fuzzy Association Rules) for ubiquitous data stream. The developed approach is stated as FFP_USTREAM (Fuzzy Frequent Pattern Ubiquitous Stream). This phase is performed in three stages such as sliding window, fuzzification and classification. Finally, the proposed FFP_USTREAM is applied in sigmoidal RNN (Recurrent Neural network) integrated with ANFIS (Adaptive Neuro Fuzzy Inference System). In the second phase, IFWIAR (Improved Fuzzy Weighted-Iterative Association Rule) is proposed for reduction of computational time with increased dataset classification accuracy. This IFWIAR uses filtration process for improving accuracy. The proposed IFWIAR adds weights to dataset for computation and optimal weights are estimated. Through estimation of weights computational time will be minimized. In the third phase, WSMS-FISW (Weighted Stream Mining Support In Frequent Items Sliding Window) is proposed for reducing run-time of frequent data patterns. This stage uses sliding window for frequent data pattern processing. By the utilization of tree structure in sliding window, number of transactions for processing is reduced. The reduced frequent data transaction leads to reduced run-time newline