الفهرس | Only 14 pages are availabe for public view |
Abstract Users are engaged to social networks because of its huge amount of services. Analyzing interactions between human and web is very important to understand human behavior on the web. Thus, a better customization could be produced according to each user’s needs and interests. Many studies produced models to explain user behaviors without clarified prediction of classes for the user expected behaviors. This research proposes a model of the social network user behavior analysis using Bayesian network (SNUABN). Proposed model is depending on determining user features and behaviors. Dataset is gathered from the Scientific Computing Center (Mansoura University) official Facebook webpage. Then, by using Bayesian Networks, performing the analysis and generating classes’ probabilities for user behaviors. The top class probability is supposed to be the predicted user behavior. Finally the proposed model is tested according to comparing the training dataset with testing dataset. Proposed model is accomplished using both Bayesian Networks and Logistic Model Tree. A comparison between the percent of the correctly classified instance of each classifier (Bayesian Networks and Logistic Model Tree) is done showing close result. And dynamism is proposed as a future work. A prediction of user behavior with a high level of accuracy is the result of the proposed model. |