ASN: A method of optimality for seed identification in the influence diffusion process

The influence phenomenon in any social network highly relies on its influential seed nodes. However, the majority of the existing research is on single-phase diffusion models where the seed nodes are chosen at once to initiate the diffusion process, and the influence diffusion is primarily investigated using progressive models. Thus, these models may not work effectively for some real-life events where the influenced users get uninfluenced in the future. Also, many existing seed selection schemes either rely on the network’s structure or relationships between nodes. Hence, these methods might not offer an optimal seed identification solution. This paper presents a non-progressive diffusion model named the ICIS model, which handles non-progressive influence diffusion across multiple time phases. This paper establishes a relation between the node’s state change in the ICIS model and the dynamics of queueing theory to analyse the influence potential of the nodes. In this paper, we also propose an optimal seed selection method named the ‘ASN’ method that considers the effects of a node’s state change to accurately compute the advantage value for each node. Thus, regardless of the topological characteristics of the network, this method offers an optimal means of choosing the seed nodes in the network. An experimental investigation on a few networks illustrates the efficiency of the proposed method. By utilizing the ASN method, we also estimate the percentage deviation of many existing seed selection techniques from the optimal solution. •Presents a non-progressive influence diffusion model named as the ICIS model.•Present an optimal seed selection scheme named as ASN method.•Experimental study on datasets demonstrate the superiority of the proposed method.