Design of viral marketing strategies for product cross-sell through social networks

Abstract

In this paper, we introduce a novel and generalized version of the influence maximization problem in social networks, which we call as Budgeted Influence Maximization with Cross-sell of Products (B-IMCP), and it considers simultaneously the following three practical aspects: (i) Often cross-sell among products is possible, (ii) Product-specific costs (and benefits) for promoting the products have to be considered, and (iii) Since a company often has budget constraints, the initial seeds have to be chosen within a given budget. In particular, we consider that the cross-sell relationships among the products of a single company are given by an arbitrary bipartite graph. We explore two variants of cross-sell, one weak and one strong, and also assume product-specific costs and benefits. This leads to two different versions of the B-IMCP problem. Given a fixed budget, one of the key issues associated with each version of the B-IMCP problem is to choose the initial seeds within this budget not only for the individual products, but also for promoting cross-sell phenomenon among these products. The following are the specific contributions of this paper: (i) We propose a novel influence propagation model to capture both the cross-sell phenomenon and the costs---benefits for the products; (ii) For each version of the B-IMCP problem, we note that the problem turns out to be NP-hard, and then, we present a simple greedy approximation algorithm for the same. We derive the approximation ratio of this greedy algorithm by drawing upon certain key results from the theory of matroids; (iii) We then outline three heuristics based on well-known concepts from the sociology literature; and (iv) Finally, we experimentally compare and contrast the proposed algorithms and heuristics using certain well-known social network data sets such as WikiVote trust network, Epinions, and Telco call detail records data. Based on the experiments, we consistently found that the stronger the cross-sell relationship between the products, the larger the overlap between the seeds of these products and lesser the distances among the corresponding non-overlapping seeds.