Hessian Distributed Ant Optimized Perron Frobenius Eigen Centrality Using Multiagent For Social Networks

Vinita Tapaskar, Mallikarjun M Math

Rising number of applications have started dealing with terabytes of data and data analysis hurdles user decision making in a timely manner. At the same time, at present, there is an increase in demand towards an inclination of big data analysis for social networks. So, one of the solution to efficient big data analysis for social networks is using distributed multi-agent-based method in an optimized manner. Examining the formation of a social network assists in acquiring comprehensions into interactions and associations between users and directs them in selecting their choices. In this work, to process huge data (i.e. Big Data) in a computational efficient and accurate manner, a method called, Hessian Distributed Ant Optimized and Perron Frobenius Eigen Centrality (HDAO-PFEC) is proposed. The main aim of HDAO-PFEC is to design adaptive Multi Agent System framework for big data analysis. The expected result with this approach is increase in accuracy of data processing around 8 to 10%, reduction in storage needs up to 10% and expectation is reduction in processing time up to 15% as compared to laplace three-level stochastic variational inference and multi-agent based distributed architecture. First, computationally efficient similar user interest tweets are obtained by means of Hessian Mutual Distributed Ant Optimization MapReduce model. Eigen Vector Centrality is a metric of significance of an influential node (i.e. social network user) in a network (i.e. social network), which permits association to other important nodes (users) and therefore having greater influence on social networks. With this objective, a distributed computing method for the estimation of Eigen Vector Centrality value for each social network user with the aid of MapReduce approach in the Hadoop platform involving Big Data which allows speedy and well organized calculations is presented. Finally, immense investigational experimental learning proves the usefulness and accuracy of HDAO-PFEC method along with the time and overhead on well-known sentiment140 dataset.

Volume 12 | 06-Special Issue

Pages: 1007-1019

DOI: 10.5373/JARDCS/V12SP6/SP20201119