An Efficient Optimization Method of Advertising on Online Social Networks Based on Data Mining


  • Tingping Zhang, Di Wan, Ren Li


How to advertise in online social networks is a hot and open research topic. In short, its goal is to post the advertisement on a few of most influential users’ profiles to spread the advertising information to the potential suitable recipients. Typical research works about this topic might be involved with two aspects: one is Spreading Maximization problem and the other is Centrality measures. The Spreading Maximization is proved to a NP-hard problem, which means the corresponding method is mostly inefficient to the topic in this paper. Second, traditional centrality measures, such as degree centrality, closeness centrality etc., roughly take the geometric information (degree, distance) to calculate the potentially most influential users, rather than considering online users’ personal interests or preference which are more likely to determine the set of people whether read/accept the advertising contents or not. In this paper, we put closed-related labels for each individual’s profile in the network and assign scores to these attribute labels. Based on these labels, we apply the weighted k-shell decomposition method to identify the core users in the networks, which is also regarded as the most influential users in this paper. The experimental results show that the proposed method is sufficient to identify the most influential users in some artificial networks. More importantly, the proposed method show good discrimination degree of influence ranking.