This project analyzes a social network, not necessarily intended as a social network like Facebook (in this case we are talking about online social networks). In this case, in fact, the network was built with a web crawler in PHP, which recursively starting from a given initial node, explores all the links coming out of the node and sequentially visits all the new nodes discovered.

1. Initial exploratory analysis

The analyzes carried out on the network thus collected were a comparison with classical models, such as the Erdos-Renyi model and the Barabasi model. To then carry out an exploration of the data by calculating the metrics related to the network, such as average distance, diameter, global clustering coefficient, etc.

2. News diffusion analysis between websites

Then the phenomenon of news diffusion between websites was analyzed using models such as the SI model (Susceptible-Infected model) and the threshold model.

3. Link prediction

The goal of this part is to randomly remove links from the network, and to use link prediction algorithms to check how many links were predicted correctly and how many were not.

4. Removal of the nodes

In this last task the resistance of the network to a hypothetical attack that would collapse the network was analyzed, removing the network nodes according to different algorithms (randomly and then according to well-defined criteria).

Download of the document

If you want to find more information, related to this project, you can find the download link below (only italian)