Kim, YanggonLee, Hyeonchoel2015-12-172015-12-172015-09-042015-05DSP2015Leehttp://hdl.handle.net/11603/2088(D.Sc.) -- Towson University, 2015.In recent years, diverse social networks, such as Facebook, YouTube, and Twitter, have rapidly grown in size and influence, and a huge amount of data is being generated from the social networks in real time. Demands for data mining on social networks have been dramatically increasing, since analyzing the data can yield insights and understanding to real world phenomenon. However, there are a lot of challenges and difficulties with data collection, management, and analysis because of the features of the social networks' data: large, noisy, and dynamic. Therefore, this study will address the overall problems with data mining in social networks and improve existing data mining techniques. We propose an integrated data collection, management and analysis system. Furthermore, we propose specific analysis methods, such as topic classification, sentiment analysis, and seed selection algorithm to analyze social networks' data.Architecture and process of data collection and management system -- Topic related document classification -- Sentiment analysis using a probability model -- Dynamic seed selection -- Case study: data analysis in the view of mass communicationapplication/pdfxiii, 104 pagesengCopyright protected, all rights reserved.Text