In the era of information explosion, social media has changed people's behavior dramatically and more and more researchers have involved in the study of social networks. Social networks provide effective marketing mediums for businesses, especially viral marketing, in which has the features of rapid propagation and low cost. This thesis presents a simulation model of viral marketing in social networks and proposes two strategies for selecting initial seed nodes, namely, multi-clique seeding and single-cluster seeding. The proposed seeding strategies are more practical than most previous works in that our decision-making for initial seeds are based only on local, rather than global, topological information of a social network.