Abstract
All current recommendation algorithms, when modeling user–item interactions, basically use dot product. This dot product calculation is derived from matrix factorization. We argue that an inherent drawback of matrix factorization is that latent semantic vectors of users or items sometimes do not satisfy triangular inequalities, which may affect the performance of the recommendation. Recently, metric factorization was proposed to replace matrix factorization and has achieved some improvements in terms of recommendation accuracy. However, similar to matrix factorization, metric factorization still uses a simple, linear fashion. In this paper, we explore leveraging nonlinear deep neural networks to realize Euclidean distance interaction between users and items. We propose a generic Neural Metric Factorization Framework (NMetricF), which learns representations for users and items by incorporating Euclidean metric factorization into deep neural networks. Extensive experiments on six real-world datasets show that, compared to the previous recommendation algorithms based purely on rating data, NMetricF achieves the best performance.
Highlights
Servers and web pages connected to the Internet show an explosive trend
In order to overcome the information failure caused by information overload, the personalized recommendation system came into being
Combining metric factorization and deep learning techniques, we propose the neural metric factorization (NMetricF) method
Summary
Massive amounts of information are presented to us at the same time. Traditional search algorithms can only present users with the same results of sorting items but can not provide corresponding services for different users’ interests and hobbies. The personalized recommendation algorithm evaluates the user’s preferences based on his online interaction history (e.g., purchase, click, browse, comment, etc.), combined with some auxiliary information (e.g., attributes, social connections, text, knowledge graph, etc.). It selects the N items from the candidate set and recommends them to the user
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have