چكيده به لاتين
The human’s life has been tied up with the artificial social networks in the modern and contemporary communities. The social networks’ users spend billions of hours on reading and sharing each other’s comments and opinions, as well as creating new contents. The present research attempts to develop a model and a framework, by which the men’s behavior towards sending and receiving messages in social networks can be explained. The present dissertation considers SIR model (one of the most renowned epidemic diffusion models) as a backbone for its proposed models. By using the idea of remembering and forgetting curve, the research introduces a novel diffusion model (namely “RbID”) in social networks literature. Then, it introduces another novel diffusion model (called “CRbID”), enlightening adoption and propagation dynamics of opinions through social networks by repeatedly calculating the nodes’ states.
Furthermore, in the present research, two separate algorithms, “GTCb” (only for SIR) and Genetic Algorithm(GA), have been developed so as to solve influence maximization problem through the complex networks. The manifold simulations which are carried out on 6 different networks, employing SIR platform, show that the GTCb’s diffusion quality outperforms all 4 well-known centrality measures as well as TOPSIS-based approach by far (by more than 8% in USAir network). However, GA has had the best quality among the proposed algorithms, in terms of the runtime, GTCb identifies its set of seed nodes 10 to 1000 times faster than GA.
The simulations, carried out on RbID and CRbID, clarify the dynamics of votes that each player secures in the networks. The results show that the message sending scheduling has a considerable impact on number of votes that each player receives, apart from the power(centrality) of selected seeds. In addition, through the simulations, using CRbID, the significance of messages’ content optimization, alongside with the message sending scheduling, has been represented.
The current research employed a real social network, so-called “C-Friends”, as an example to explain CRbID. In this example, a cellphone application which had won only 5.5% of votes, compared to its rival, has managed to extend its own votes to 22% averagely by sending only 2 messages to the network, by implementation of the proposed GA.
