Auxiliary Information-Enhanced Recommendations

Sequential recommendations have attracted increasing attention from both academia and industry in recent years. They predict a given user’s next choice of items by mainly modeling the sequential relations over sequence interactions with items. However, most existing recommendation algorithms focus on dependencies between item IDs within sequences, while ignoring rich complex embedded auxiliary information, such as items’ image information textual information. Such can help us better understand users’ preferences towards items, thus benefit recommendations. To bridge this gap, we propose an information-enhanced algorithm called memory fusion network for (MFN4Rec) to incorporate Accordingly, IDs, are regarded three modalities. By comprehensively modelling modalities interaction across modalities, MFN4Rec learn more informative representation accurate Extensive experiments two real-world datasets demonstrate superiority state-of-the-art algorithms.