Attribute-Based Influence Maximization in Social Networks

As traditional advertising model exposes its weakness of ignoring consumer interests, the concept of narrow advertising draws increasingly more attention which considers the feature of each user. Under this specific environment, effective viral marketing has to select a set of initial users to maximize their influence on the targeted customers. This paper aims at the integration of viral marketing and narrow advertising, by proposing a novel problem called attribute-based influence maximization. Firstly, the problem definition is presented with the consideration of user features. Then the influence probability between two nodes is modeled and two heuristic algorithms, Sum of Probability Covered Algorithm (SoPCA) and Community-based Algorithm (CBA), are designed. Finally, experiments on six datasets are conducted to verify the effectiveness of proposed algorithms.