Graph Ontology Based IAFCM Model for Itemset Mining in Transactional Database: An Optimized Fuzzy Framework

Abstract

HUIM (High Utility Itemset Mining) is one of the most analyzed data mining activities. Product suggestion, e-learning, bioinformatics, text mining, market basket analysis, and web click stream analysis are few of the areas where it can be used. Cost savings, greater competitive advantage, and increased revenue are the advantages gained by pattern analysis. However, because HUIM approaches do not examine the correlation of retrieved patterns, they may uncover false patterns. As a result, a number of technique for mining related HUIs have been presented. These algorithms still have issues with computational cost, both in conditions of period and memory usage. As a result, a method for mining weighted temporal patterns is proposed. The suggested method begins by preprocessing time series-based data into fuzzy itemsets. These are fed into the Improved Adaptive Fuzzy C Means (IAFCM) technique, which is a hybrid of the FCM clustering method and the Graph based Ant Colony Optimization (GACO) technique. The proposed IAFCM technique accomplishes two goals: IAFCM clustering and data reduction in FCM clusters, and ii) optimal itemet placement in clusters using GACO. Using GACO, the suggested technique produces high-quality clusters. On these clusters, weighted sequential pattern mining is used to find the most effective sequential patterns, which take into account knowledge of patterns with low frequency and high weight in a repository that is updated over time. The results of this method show that when compared to other traditional methodologies, the IAFCM with GACO improves execution time. Furthermore, it improves the data representation process by increasing accuracy while using less memory.