Mining the network value of customers
Proceedings of the seventh ACM SIGKDD international conference on Knowledge discovery and data mining
Maximizing the spread of influence through a social network
Proceedings of the ninth ACM SIGKDD international conference on Knowledge discovery and data mining
Cost-effective outbreak detection in networks
Proceedings of the 13th ACM SIGKDD international conference on Knowledge discovery and data mining
Efficient influence maximization in social networks
Proceedings of the 15th ACM SIGKDD international conference on Knowledge discovery and data mining
Everyone's an influencer: quantifying influence on twitter
Proceedings of the fourth ACM international conference on Web search and data mining
Influence and passivity in social media
Proceedings of the 20th international conference companion on World wide web
| Hi-index | 0.00 |
Identifying the most influential nodes in a network is a well known problem which has received significant attention in the research community. However, as networks grow larger, an interesting variation of the problem becomes relevant where in the aim is to maximize the influence not in the whole network but only on a sub-set of the nodes in the network. We approach the subset specific top-k influential problem standalone and show that, unlike traditional approaches, a search for subset specific top-k influentials can be terminated earlier based on a parameter γ - thus allowing a trade-off between efficiency and effectiveness. For social networks, this parameter has a behavioral interpretation: it captures the ease of influencing nodes in the network. This work makes three key contributions. First, we propose an iterative network pruning algorithm to find subset specific top-k influentials and compare its performance to subset-adapted existing algorithms for various values of γ on real world data sets. Second, we extend the existing analytical framework for top-k influential detection to incorporate γ. Third, we analyze our algorithm under our analytical framework and show that the influence spread function continues to be sub-modular. Though our work has been motivated by online social networks, we believe that it is useful in other domains where diffusion over networks is considered.