Libro

Descripción

In this research, some of the challenges of social networks such as fake news detection, link prediction, recommender system optimizations, information dissemination prediction, and data forwarding are studied. In chapter 1 a new method is proposed to improve the fake news classification performance by reducing the data size using the feature selection techniques so that a feature subset is chosen from the original set. In Chapter 2, a new link prediction technique is introduced which uses the Louvain algorithm to cluster nodes and then estimate future links by analysing the node relationships. Chapter 3 presents a new method for recommender systems that is based on information dissemination and popularity on social networks. This method improves the issues and drawbacks of the previous methods such as forecast accuracy and coverage. Chapters 4 and 5 explore 2 new methods for predicting information dissemination patterns. In the first method, we use the Laplacian centrality criterion and ant colony algorithm to identify the optimal propagation path. In the second method, it is tried to improve the first method by leveraging the densest subgraph algorithm to create a subset of maximum independent nodes. Finally, in chapter 6, a new approach is presented to improve the time and accuracy of data distribution on social networks.