Indian Journal of Science and Technology

Abstract

Background/Objective: With the development of data mining techniques to analyze large amount of complex data has played an essential role in several areas like cloud computing, medical databases, geographical information retrieval, etc. The automatic evaluation of cloud patterns is a challenging task due to the large amount of interesting patterns can be extracted. However, how to find infrequent patterns is still an open issue in cloud computing. Methods: Conventional approaches are mainly depends on quantitative datasets with support and confidence measures. Due to the large amount of cloud storage data, it is very difficult to extract the weighted association rules based on the server usage statistics. Traditional techniques are implemented on the data samples with the same attribute type. Due to this fact, a multi-class algorithm is proposed to find relevant usage patterns of large complex data. Findings: Proposed approach does not rely on any probabilistic closure measures and quantitative data. This approach minimizes the database scans and optimizes the infrequent cloud patterns. Applications/Improvements: Experimental results show that, proposed work generates high quality cloud patterns compared to traditional quantitative rule mining techniques. 
Keywords: Complex Cloud Data, Infrequent Association Rules, Multiclass Attributes