Knowledge Discovery in Big Data

Abstract

Big Data is quickly attaining impetus thereby inviting community of researchers and organization from all the sectors to explore its tremendous potential. Big Data is deliberated to be probable raw material helping to attain domain specific knowledge for gaining comprehensions related to managing, planning, predicting and security etc. Big Data era comes with challenge to explore not just data the right data and using computers to extend our domain knowledge by identifying patterns that we did not see or could not find previously. newlineUnited with the Knowledge Discovery process, Big Data movement unleash tremendous distinctive prospects for organizations to gain benefits by excavating knowledge. However, keeping in consideration the difficulty to analyze massive datasets, unique architectural and systems engineering challenges are presented by Big Data. The challenge of analyzing Big Data lies in dealing with great quantity, exhaustiveness and variability, timeliness and dynamism, disorderliness and ambiguity, high relationality, and the fact that most of the generated data does not focus on s specific question to be answered or is an outcome of another task. newlineThis thesis addresses issues related to Big Data Analytics architectures and models. Based on the careful examination of these issues, this work proposes three models for Big Data Knowledge Discovery namely, KDBDA: Knowledge Discovery model for Big Data Analytics, Service-Oriented model for Big Data Knowledge Discovery and and#956;BIGMSA- Microservice based model for Big Data Knowledge Discovery. newlineIn order to overcome issues related to accuracy, quality and efficiency related to machine learning algorithm when Big Data is concerned, a fusion algorithm for Big Data Pre-processing ACO-clustering algorithm is proposed. The suggested algorithm will improve and escalate the search speed by optimizing the process. As the projected method use ant colony optimization along with clustering algorithm it contributes in decreasing pre-processing time and enhancing the