Data of social media can often be readily collected from social networking sites and blogging sites with site-provided APIs. What is different from classic sociology data collection for research is that abundant data often exist, but cannot be directly used without extensive data processing such as information retrieval and text mining. This is because traditional questionnaire-based methods do not work well partly due to user ID anonymity. In order to conduct the proposed comparative study, therefore, we need to design and develop special research tools that enable us to retrieve research related data, summarize and organize them into forms that are compatible with classic sociology research investigation. We thus propose to focus on the investigation of some key research issues about blogosphere: (i) determining group detection, profiling and evolution, (ii) summarizing and understanding what is going on by extracting events of interest, (iii) performing sentiment analysis and opinion mining, and (iv) analyzing information diffusion and influence propagation as well as tracking users across blogging sites or social networking sites. We will investigate events occurring in the region through content analysis of news media stories about specific occurrences, making use of established content analysis methods for determining the parties involved and the actions each party undertook, including violent action events. We elaborate the above four research issues with specific technical details in the proposal.
|Effective start/end date||6/1/09 → 12/31/11|
- DOD-NAVY: Office of Naval Research (ONR): $284,721.00