|
Abstract
|
Influence maximization techniques emphasize
selecting a set of influential nodes in order to maximize influence.
Because the algorithms presented in this field ignore the topology
of cliques for diffusion, there are two major challenges in influence
maximization algorithms: optimal diffusion and computational
overhead reduction. As a result, the CDP algorithm is presented
in this article to address these issues. This algorithm first selects
suitable cliques for diffusion based on their position and strategy
in social networks. Furthermore, for each clique, a score is
calculated based on the topological criteria of the clique, and
suitable cliques are selected for diffusion by applying a threshold
limit. The seed nodes are chosen from the cliques in the second
step. To avoid the rich club phenomenon, only a few nodes from
each clique are chosen as seed candidate nodes. Finally, the seed
nodes are chosen based on the node's topology and the strength of
the node's level one neighbor. In the experiment section, the CDP
algorithm significantly outperforms the best algorithms presented
in recent years in terms of influence spread rate and execution
time.
|