User profiling and recommender prototype

This project was a partnership between Queensland University of Technology (QUT) and the Fairfax under the umbrella of CRCSS (CRC – Smart Services) with a focus on building new personalised services based on data mining. The objective was to build a personalised system that will enable more intuitive search for specific products and relevant advertisements by accommodating recommendations made by user profile and product profile models. It includes the development of user profiles by mining user search and click log data, development of product models from users’ perspectives by mining user online product reviews, and develop predictive models that can predict users’ product preferences based on user and product profiles and recommend products using the predicted user’s preferences. This project charted new ground due to the proposal of reciprocal recommendation. We have developed a number of web personalisation methods for product recommendation as well as people-to-people recommendation in online social networks.