Link Prediction

In this chapter, the approaches to link prediction based on measures for analyzing the 'proximity' of vertices in a network have been worked upon. The notion of closeness is defined by a value called the 'proximity measure'. Several distance or proximity measures such as graph distance, common neighbours, Jaccard, cosine, Adamic/Adar, preferential attachment and many more are described with proper exemplification to compute proximity measures for link prediction. This chapter describes the approaches to link prediction based on measures for analyzing the 'proximity' of vertices in a network have been worked upon. Social networks are a popular way to model interactions among people in a group or community. The measure follows the notion that social networks are small worlds, in which individuals are related through short chains. The common neighbours predictor captures the notion that two strangers who have a common friend may be introduced by that friend. The Adamic/Adar predictor formalizes the intuitive notion that rare features are more telling; documents that share the phrase 'for example' are probably less similar than documents that share the phrase 'clustering coefficient'. One well-known concept in social networks is that users with many friends tend to create more connections in the future.