Mining Facebook Friendship Networks
Without a doubt, Facebook is one of the most important social networking sites to date. Since its launch in 2004, its user friendly environment and interface led to its rapid development, transforming into a platform able to connect billions of people in a integrated mesh of “friendships.”
A recently published research paper, “Using Mining Predict Relationships on Social Media Network: Facebook,” (https://thesai.org/Downloads/IJARAI/Volume4No4/Paper_9-Using_Mining_Predict_Relationships_on_the_Social_Media_Network.pdf), looks into the notions of friendships and relationships on such a network. Most of the paper mainly covers the sampling algorithms used to extract data from the site. It discusses the detailed procedural analysis it used to conduct an anatomic network study of not only Facebook, but other online social networks as well. In order to sample and analyze data, the researchers used a web mining architecture called a crawler agent, which would collect data that could be used to compare and analyze quality and behavioral properties. This would all be mapped into graphical structures that allowed the researchers to easier understand structural traits and behaviors.
This data mining methodology delves deeper in the prediction and analysis of social connections than the simplification of these ties the triadic closure generalizes. Rather than narrowing the network view to a small cluster of nodes, crawler agent is able to extract data from an enormous network that people would not be capable of performing with the speed and precision of this platform. As impressive as this web mining architecture is, it is severely limited in that it only contains data from the web and cannot confide any data from outside social networks. People interact and form new connections everyday outside of social media. Some of these connections might be tied to those who do not actively or ever participate in social web networking. Strong and weak ties, in some incidences, may not even be distinguished accurately due to data out of reach for web mining and the crawler agent. In spite of some shortcomings on the bigger picture, the sheer volume of data collected and the accuracy of the predictions it outputs is a feat to behold.