In recent years, a large number of social network sites have appeared to connect people and groups together. Networks have been proven to be a good tool to share information and communicate ideas. Influence propagation occurs when an individual’s opinions or behaviors change as a result of interactions with others. The influence maximization problem aims to identify a subset of initial adopters in a social network to maximize the influence propagation. There are two progressive models most used in the analysis of social networks, namely the independent cascade model and the linear threshold model. As a type of epidemic models, the independent cascade model assumes that an individual adopts an innovationwith a certain probability if at least one of its in-neighbors has adopted it. Differently, the linear threshold model assumes that an individual adopts an innovation if a certain ratio of its in-neighbors have already adopted it. The thesis addresses three problems: influence propagation computation, influence maximization by seed selection and influence maximization by link activation. The influence propagation computationconsist in computing the probability that each node can be activated given a certain set of initial adopters. We propose the PathMethod to give an exact result, the SSS-Noself algorithm and the SSS-Bounded-Path algorithm to give an approximate result. The influence maximization by seed selection consist in maximizing the final influence propagation by targeting a seed set of certain cardinality. We initially propose the problem of influence maximization by link activation. Various properties of this problem and some sub-optimal solutions are given