Multi-Sources Unstructured Data Analytics
This project is a continuation of research from NExT’s Phase I on big unstructured data analytics. The aim is to leverage our research results in Phase I to analyse and fuse multi-modal data arising from multiple data sources including social media, Web, forums, and, if available, structured and proprietary data sources. The research will target applications in the domains of Smart City, Fintech and E-Commerce.
The primary difference in this phase is to attain more accurate and complete extraction and discovery of knowledge from multiple unstructured data sources. This means that we need to develop better technologies for the aggregation of multi-source data streams; understanding of text, video and sensory contents; removal noise and spams; extraction of key evolving topics and trends; profiling of users and communities; and generation of predictive and prescription analytics. For the first 2 years, the aim towards applications is the inference of (actionable) insights in different domains, and the discovery of online falsehood. Accordingly, we will target our research on (semi) automated techniques to gather data and constructing knowledge graph; as well as various multi-source data integration techniques.