Grant: Collaborative Research and Development Grants

This project focuses on developing techniques within a unique platform for product knowledge base development, management and use. Our work is based on the following assumptions: i) The Web is a rich environment for extracting product information as a result of recent initiatives by large retailers such as Sears, who are now embedding rich structured Microdata of their products in their product pages; ii) Linking of product descriptions with concepts from Linked Open Data is able to enhance the process of finding similarities between products, making meaningful content-based recommendations possible; iii) Semantically annotating product data can effectively assist in the process of product knowledge base consolidation; and iv) The collective behavior of the users, observed as a result of their interaction with the recommendations, can be seen as an indirect measure of the quality of the recommender algorithms and the underlying product knowledge base, and could potentially be used as a means to enhance the knowledge base and the recommendation algorithms. Based on the assumptions, three objectives have been defined for this project:1) Develop suitable algorithms that would aggregate structured product data from various sources over the Web, annotate the extracted product information with concepts from ontologies of the Linked Open Data, semantically match similar products and finally build a unique integrated product knowledge base;2) Devise appropriate content-based recommendation algorithms that consider the semantic information represented in the product knowledge base to make meaningful suggestions. Semantic and information retrieval-based measures for computing product similarity in the product knowledge base need to be devised;3) Create a continuous feedback loop between the product knowledge base and the recommendation algorithm in a way that the implicit feedback from the users on the recommendations would be considered as a measure of the suitability of the product knowledge base and the performance of the recommendation algorithms.