User Profiling with publicity Available social media data

Abstract

The expansion of the social media usage we witnessed in the recent past has sig- ni cantly contributed to the personal information exposure in today's world. Due to the ability for anyone to add data on anybody and accessibility to the enormous amount of personal information, combined with the advancement of INTERNET search engines, people can easily nd data on others. However it can be di cult to infer meaningful information due to the vast amount of unorganized, available data. Social media sites naturally contain large amounts of data and signi cant amount of this data tend to be less informative when they viewed separately. As a solution for this we propose an approach based on Text Mining and Natural Language Processing techniques to automate this information extraction process. Our approach collect information from social media feeds and build a secondary pro le for a subjected user. To test our approach we used data from micro blogging social media platform Twitter. We chose Twitter because it has a big, active user base and by default Twitter data is public. When compared to other social media networks, Twitter provides easy and unrestricted access to the data. We evaluated our approach by building pro les for set of random users. As the results suggests our proposed approach can be used for constructing a pro le for any person with a digital foot print of fair size. Also as the results suggest this approach can be applied in many domains such as information aggregation, moni- toring privacy leaks, monitoring suspect activities and such. Even though we only used the data from Twitter, the proposed approach can be expanded to use with multiple data sources. Which would aggregate scattered data on speci ed users to build information rich user pro les.