Mining relationships from text in social networking sites

Abstract

Majority of an individual's social interaction takes place through social networking site like Facebook. But one thing to focus is that the way we interact with different people through these social networking sites are different. If we are interacting with our close friends we use more informal language whereas with our relatives or elders and with our seniors at job interaction is little less informal or some time formal also as compared to the interaction with close friends. This difference in interaction can be a good measure of guessing relationship of the persons interacting with each other through these social networking sites. In this research paper we collected data through one of the very famous and very fast growing social networking site, Facebook. We considered comments, wall posts and messages exchanged through chat boxes of some people. From this data we have extracted some very interesting features like difference in number of emoticons, type of emoticons, degree of informal language, degree of intentional spelling mistakes, frequency of social acronyms and degree and type of interjections used in interaction of different people. We have used these features for mining relationships of the people involved in online social interaction.