The paper addresses the problem of selecting a set of influential nodes in a social network to maximize the spread of influence, motivated by applications in viral marketing. The authors consider two widely studied models: the Independent Cascade Model and the Linear Threshold Model. They prove that the influence function in both models is submodular, which allows them to derive approximation guarantees for a greedy algorithm that selects nodes based on their marginal gain. The greedy algorithm is shown to achieve a solution within 63% of the optimal in several classes of models. Computational experiments on large collaboration networks demonstrate that the greedy algorithm outperforms node-selection heuristics based on degree and distance centrality. The paper also introduces a general framework for diffusion processes in social networks, unifying the Linear Threshold and Independent Cascade models, and discusses extensions to more complex marketing actions and non-progressive processes.The paper addresses the problem of selecting a set of influential nodes in a social network to maximize the spread of influence, motivated by applications in viral marketing. The authors consider two widely studied models: the Independent Cascade Model and the Linear Threshold Model. They prove that the influence function in both models is submodular, which allows them to derive approximation guarantees for a greedy algorithm that selects nodes based on their marginal gain. The greedy algorithm is shown to achieve a solution within 63% of the optimal in several classes of models. Computational experiments on large collaboration networks demonstrate that the greedy algorithm outperforms node-selection heuristics based on degree and distance centrality. The paper also introduces a general framework for diffusion processes in social networks, unifying the Linear Threshold and Independent Cascade models, and discusses extensions to more complex marketing actions and non-progressive processes.