Identification and classification of best spreader in the domain of interest over the social networks

The emerging social networks promptly create greater opportunities for fast-developing viral marketing. The online social networks (OSNs) play an essential role in the information diffusion among the social users within the community. The social network being large-scale, it leads to the inconvenience in identifying the influential spreaders in a specific domain, as every social user receives the information from different sources through multiple connections over the network. Although, analyzing the complex social network is indispensable to determine the influence spreaders with the knowledge of understanding the dynamics of information evolution. The existing solutions of the influential measurement techniques lack in neglecting the redundant links and quantifying the temporal information among the social users while estimating the diffusion importance of a social user. Moreover, these techniques fail in analyzing the structural relationships in the domain. To overcome these obstacles, this paper presents a de-duplicated k-shell influence estimation (DKIE) model in the social network by classifying the influential spreaders based on the domain of interest using k-shell decomposition and N-gram similarity. The DKIE model incorporates two major phases such as generic influential spreader identification and domain-specific influential spreader identification. The first phase measures the diffusion importance of each active social user based on the structural relationships of the social network using k-shell decomposition method. It separates the core-like groups and true core and identifies the best spreaders regardless of the redundant links. The second phase exploits the topic of the discussion of the best spreaders and consequently, measures the topic-wise influence to categorize the domain-specific best spreaders using N-gram similarity measurement. The experimental results illustrate the effectiveness of DKIE approach.