Research
Generally, our platform aims to provide users two key functionalities:
- Entity-Centric Data Monitoring for Focused Interest: Our platform supports creating a “relevant” social media data corpus for any analytic application by effective data monitoring. It is unnecessary and inefficient to collect the entire social media data. Therefore, our platform should allow users to easily define “relevance” for an application based on the entities of interest, and automatically collects only relevant documents (and users) for those entities, as the corpus to analyze, from cloud servers to local computers.
- Users Profiling to Enable User-Aware Analytics: Our platform supports doing analytics over a specific segment of the collected social media data w.r.t. the social media user’s categories by effective user profiling. Entity-centric analysis usually requires being of aware of different attributes of users (e.g., age group and education level). Therefore, our platform should have a way to learn social users’ attribute values, , and allow people to flexibly choose the appropriate segment of the user pool for user-aware analytics.
Given these two key functioalities, we focus on the following three main concepts in our research for BigSocial project:
- Accessible Data: Getting the social data for topic/entity of interest.
- Attributable Users: Knowing who is speaking; where the signals come.
- Accountable Social Phenomena: Finding what users and what community hold certain social phenomena.