Multi objective functional frequent subgraph mining using affinity measurement on mapreduce

Abstract

Data mining refers to extracting and analysing meaningful information from large volume of data. The datasets such as healthcare, financial banking, Fraud detection system, intrusion detection system and customer feedback dataset may contain information and relation between data components can provide the information through algorithms. In medical dataset, the different cancer cell such as ovarian cancer, breast cancer, kidney cancer and liver cancer consists of different chemical components. Traditionally, the type of cancer cells are determined through sequential process such as data cleaning, data integration, data reduction, transformation, pattern evaluation and knowledge representation from dataset. The above data mining process have shortcomings such as, requirement of skilled professional to analyse, validate data and during data mining, the information gathering process can be overwhelming, where more information are gather from dataset. Hence, the data mining process time and the amount of relevant information gathering from dataset improve through subgraph mining algorithm. The subgraph mining discovers sequential and non-sequential patterns of data present in dataset. Traditional mining algorithms apply for different domains of dataset to analyse geometric, spatial and topological relation between data in terms of vertices and edges. The arbitrary relation among vertices and edges in complex dataset solves by assigning label to vertex and relational label to edge in graph. The subgraph mining helps to find recurrent substructures in graph. However, the subgraph mining requires high computational requirement. In this thesis, subgraph mining extends with newline