This paper presents an algorithm capable of providing meaningful and diversified product recommendations to small sets of users. The proposed approach works on a high-dimensional space of latent factors discovered by the bias-SVD matrix factorization techniques. While latent factor models have been widely used for single users, in this paper we formalize recommendations for multi-user as a multi-objective minimization problem. In the pursuit of recommendation diversity, we introduce a metric that explores the angles among product factor vectors and extracts from these a measurable real-life meaning in terms of diversity. In contrast to the majority of recommender systems for groups described in literature, our system employs a collaborative filtering approach based on latent factor space instead of content-based or ratings merging approaches.
Dias, P., & Magalhães, J. M. D. C. (2014). Multi-user Diverse Recommendations through Greedy Vertex-Angle Maximization. In Lecture Notes in Computer Science (pp. 96-107) https://doi.org/10.1007/978-3-319-12571-8_9