Effective Recommendations based on Multi-Source Data (2014–2017)
Large-scale data collected from multiple sources such as the Web, sensor networks, academic publications, and social networks provide a new opportunity to exploit useful information for effective and efficient recommendations and decision making. In this project, we propose a new framework of recommender systems that is based on analysing relationships between different types of objects from multiple data sources. A graph model will be built to represent the extracted semantic relationships and novel linkage-analysis based algorithms will be developed for ranking objects. The results from this project will underpin many critical applications such as healthcare.