Predicting our Friends?
With the advent of the Web 2.0, and the advances in computing power, the internet has proliferated all corners of our lives. While some academics and enthusiasts point out signs of a second bubble that might soon implode on itself, one fact stands out- social networks have increasingly dominated the information technology scene and have rapidly become an integral part of our daily routine. The interesting thing about the explosion of social networks though, is the impact it has had on the exploration of new branches of mathematics and sociology. Academics have been able to adapt concepts from graph theory, game theory and combinatorics to explain why social networks behave in the manner they do, all the time making new discoveries that explain in part how the human race interacts with itself as well as predicting how it will interact in the future.
The attached publication, dealing with social networks, tries to establish a method to predict future links between people in a network. The study looks at a snapshot of an existing social network and tries to predict what new links will form in the next time frame based on measures for analyzing the “proximity” of nodes in the network. The authors (one of whom is our professor, John Klienberg) describe the network as a graph G=< V, E> (a graph containing a set of vertices V and edges E) and assign to it two time intervals t0 and t1 which are the “test” intervals and “training” intervals. The authors employ a variety of methods to test their predictions including methods based on node neighbors and methods based on calculating the shortest paths between nodes and the results show that employing these methods is about 40 to 50 times stronger than using the random operator. This shows that there is definitely some useful information contained in the network topology alone.
It is interesting to note how social networks inadvertently spurred the interests of academics and consumers alike, which only reinforces their importance to our world. Not only that, the dynamic nature of social networks implies that it is constantly evolving making the possibilities for academic research endless which in turn will help explain in greater depth the complexities of the human species.
Reference Link: http://dm.kaist.ac.kr/kse625/resources/Liben-Nowell_2007.pdf