Adaptive multi-feature budgeted profit maximization in social networks

Online social network has been one of the most important platforms for viral marketing. Most existing researches about diffusion adoptions new products on networks are diffusion. That is, only piece information product is spread network. However, in fact, may have multiple features and different independently When a user would like to purchase product, he consider all comprehensively not just one. Based this, we propose novel problem, multi-feature budgeted profit maximization (MBPM) which first considers under propagation product. Given with each node having an activation cost profit, MBPM problem seeks seed set expected no more than budget make total as large possible. We adaptive setting, where seeds chosen iteratively next selected according current results. study two models, oracle model noise model. The assumes conditional marginal any could be obtained O(1) time (1-1/e) approximation policy proposed. Under model, estimate by modifying EPIC algorithm efficient policy, return (1-exp({\epsilon}-1)) ratio. Several experiments conducted six realistic datasets compare our proposed policies their corresponding non-adaptive algorithms some heuristic policies. Experimental results show efficiencies superiorities