Abstract
A remarkable success in recommendations has been achieved by using methods based on metric learning, especially in digital marketing. However, the existing methods do not consider the relative preferences among items that users like. To overcome this issue, we propose an improved recommender model. First, the model analyses the user-item bipartite graph from historical interactions, and collects user-item similarities based on the topological features from this graph. Then, similar to other metric-based methods, both users and items are embedded as latent positions in a low-dimensional space, where users’ preferences on items are modelled as distances. Thus, we propose an improved metric-based recommender, i.e. the Graph Embedded Metric Factorisation recommender, under the assumptions that (1) the distance between a target user and an interacted-with item is determined by their topological similarity, and (2) for a target user, non-interacted items are located farther away than interacted-with ones. Comprehensive experiments on three practical datasets were implemented. Empirical results indicate that our improved recommender outperforms current state-of-the-art methods when making personalised recommendations based on users’ implicit feedback.
Highlights
Recent years have witnessed a great development of the recommendation systems (RS), especially in the digital marketing field
Interacted-with and non-interacted items are modelled in different manners: (1) for the interacted-with items, we are aware of relative preferences among them and provide a strategy that sets distances between users and interacted-with items according to the similarities from the user-item bipartite graph; (2) for the non-interacted items, they are assumed to locate farther away from target users compared to interacted-with items in the latent space
THE METRIC-BASED RECOMMENDER To overcome the disadvantages of matrix factorisation (MF)-based methods and tap into the power of metric learning, we propose a novel recommendation model, Graph Embedded Metric Factorisation (GEMF)
Summary
Recent years have witnessed a great development of the recommendation systems (RS), especially in the digital marketing field. They assume both users and items as latent vectors in a dense space, and use their dot product as a preference approximation [1] They fail for the recommendation tasks based on users’ implicit feedback. To overcome this issue, Hu employed confidential weights for both observed and missing user-item pairs and proposed a Weighted Regularized Matrix Factorisation (WRMF) method [3]. The model’s rational is to learn users’ relative preferences among interacted-with items based on the user-item bipartite graph and consider the issue from both point-wise and pair-wise perspectives.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
