Affiliations: Department of Computer Engineering, S.V. National Institute of Technology, Surat, India
Correspondence:
[*]
Corresponding author. Bharat Tidke, S.V. National Institute of Technology, Surat, India. E-mail: batidke@gmail.com.
Abstract: Social networks helps to build relationships where two or more concepts, objects, or people are connected, or in state of being connected which is multidimensional and dynamic in nature. The interactive aspect of information extraction in online social networks instigates from considerations of different parameters which levied to the invention of new metrics. These metrics normally based on their ability to adapt to existing positioning or ranking indicator approaches with intent on activities and relationships among users in modern online social network which evolves with time. Existing work on network topology analysis is mainly focused on acquiring global properties such as interactions on either synthetic network or real world data provided by some authors without involving actual scenario of social network data. This research mainly focus on supervised learning using localize properties of known influential user in terms of links evolving from online social networks data. In addition, capture top K real time influential users from the evolving social network graph of known influential user. To achieve this, we propose two approaches, first an optimal Weight based Evolving Friends Follower Ranking (WEFFR) influence ranking algorithm to assign weights by capturing adaptive degree of relationship and secondly we combine WEFFR algorithm with Page Rank algorithm (WEEFRPR) to measures influence of nodes using reciprocal influence. The experiment results on Twitter network of known influential users shows that proposed approach performs better as compare to well-known existing approaches.
Keywords: Social networks, evolving topology, influence, ranking
