ADSC’s HCCS program has envisioned human-centered systems that consist of sensing (i.e., acquisition of sensor measurements by observing a target of interest) and analytics (e.g., extracting relevant information and metadata from sensor measurements, and fusing and analyzing such information from multiple sources to derive knowledge that forms the basis of decisions and actions). This proposal aims to exploit the vast “human sensors” now prevalent in our digital society—i.e., the social media– and enable social analytics over such data, towards a human-centered cyber system.


When analyzing entity-centric social media data, we often face two problems:

  1. As the data are big, how can we monitor the data for anything relevant to the entity of interest for analytics?
    E.g., in market research, we may be interested in tweets that talk about “NEWater” (a product entity); in social analysis, we may try to monitor the of the policies of “HDB” (an organization entity).
  2. As the data are generated by users of varying demographic and societal attributes, how can we associate the entity-centric analytics w.r.t. different user attributes?
    E.g., in market research, we may want to know the age groups and education levels of popular users who talk about “NEWater”; in social analysis, we may want to understand what the age and income groups of citizens are who thought negatively about HDB.

To fulfill the vision of enabling such entity-centric analytics, we propose to design and develop BigSocial, a general social media data analytic platform. It aims to support two key functionalities:

  1. Monitoring social data relevant to real world entities of interest;
  2. Profiling social users to enable rich demographic targeting and segmentation.