710258Statistical Relational Inference for Ecommerce Recommendation TechnologyClosedGRD Proof of ConceptInnovate UKThis research project aims to demonstrate a proof of concept of a radical new approach to recommendation technology for ecommerce based on Statistical Relational Inference. Online shoppers are increasingly expecting a seamless experience no matter how they are buying and retailers are expected to deliver. Yet, current tools are time and again struggle to extract useful information or actions. This 3rdgeneration of recommendation technology will provide a framework in which the full variety of data available can be modelled; offline (training) and online (recommendation) algorithms can be applied to subsets of the data model according to algorithm capability; marketing and merchandising guidance can be applied within and between model components; and these capabilities can be flexibly combined by modelling scientists to construct specific recommendation engines for specific applications or customers. It is a qualitative advance over current, second-generation recommendation approaches. In order to minimise the technical and technical risks identified the project follows a real useroriented approach to select client-champions to test the new technologies in close collaboration with them. These client champions are leaders in their respective fields and looking to push the boundaries of what the technology can do to enhance relationships on site. The deliverable from this project will demonstrate the recommendation technology for ecommerce, together with an extensive analysis to confirm the potential for development into a viable commercial product with significant potential for UK exports over the next five years.