CONTENT2VEC: SPECIALIZING JOINT REPRESENTATIONS OF PRODUCT IMAGES AND TEXT FOR THE TASK OF PRODUCT RECOMMENDATION

We propose a unified product embedded representation that is optimized for the task of retrieval-based product recommendation. We generate this representation using Content2Vec, a new deep architecture that merges product content infor- mation such as text and image and we analyze its performance on hard recom- mendation setups such as cold-start and cross-category recommendations. In the case of a normal recommendation regime where collaborative information signal is available we merge the product co-occurence information and propose a sec- ond architecture Content2vec+ and show its lift in performance versus non-hybrid approaches.