الفهرس | Only 14 pages are availabe for public view |
Abstract Information diffusion plays a significant role in online social networks, mining the latent information produced became crucial to understand how information is disseminated. It can be used for market prediction, rumor controlling, and opinion monitoring among other things. Thus, in this thesis, an information diffusion model based on dynamic individual interest is proposed. The basic idea of this model is to construct a dynamic user profile that is composed of three main features which represent user interest, behavior and influence respectively. User interest is formulated based on extracting effective topic of interest of each user overtime and identify the most relevant topics with respect to seed users. While user behavior on the social network is constructed to understand how she/he interacts with content and the frequency. Finally, the user{u2019}s position in the network and how she/he influences the neighboring users is also calculated. The information diffusion model based on the Page Rank algorithm is able to identify seed users according to the predicted information dissemination score which allows the ranking of those seeds according to the likelihood of impact. A set of experiments on real twitter dataset were conducted on each part separately (Interest, behavior and Influence) then the output of each of the three models was combined into one experiment on the proposed Information diffusion model. The experiments showed that the proposed dynamic prediction model which applies machine learning techniques accurately predicted the topics that the user is likely to interact with in the next time slice. The information diffusion model was able to rank the users of the network according to the likelihood of disseminating a certain topic in a specific time slice which would yield the highest influence on the neighboring node |