A Data mining study of ranking within social networks

Abstract

Social networks have become very popular in the past few years and have become a significant part of our personal and professional lives. As the number of participants in social networks has grown, they have become a virtual space for exerting influence. Studies in sociology and marketing have stressed the vital role of influence for businesses to survive; organizations and businesses are constantly seeking to establish and expand their presence by exploiting social networks. This has led to an implicit competition for higher visibility and ranking within social networks. Ideally, the ranking of participants within social networks should mirror the real world. However, this may or may not be true because of different degrees of participation, overrepresentation due to self-promotion and the possibility of unreliable or false information. This thesis addresses the following related questions. What are appropriate measures for ranking participants in social networks? Does the ranking within networks mirror those based on traditional measures for ranking organizations? We use data mining and statistical analysis to evaluate several measures, including a new measure based on the H-index, for ranking participants within social networks against established benchmarks for university programs. We find that prominence within social networks correlates in general with prominence in the real world. We identify the best measures for predicting prominence in the real world, and perform preliminary outlier analysis. While the observations are not proven to be causal, they offer insights of potential value to social network marketing