Collaborative filtering embeddings for memory-based recommender systems. (October 2019)

• Record Type: Journal Article
• Authors: Valcarce, Daniel; Landin, Alfonso; Parapar, Javier; Barreiro, Álvaro
• Is Part Of: Engineering applications of artificial intelligence. Volume 85(2019)
• Page Start: 347
• Page End: 356
• Publication Date: 2019-10
• Subjects: Embedding vector -- User representation -- Item representation -- Collaborative filtering -- Recommender systems; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2019.06.020
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Word embeddings techniques have attracted a lot of attention recently due to their effectiveness in different tasks. Inspired by the continuous bag-of-words model, we presentprefs2vec, a novel embedding representation of users and items for memory-based recommender systems that rely solely on user–item preferences such as ratings. To improve the performance and prevent overfitting, we use a variant of dropout as regularization, which can leverage existentword2vec implementations. Additionally, we propose a procedure for incremental learning of embeddings that boosts the applicability of our proposal to production scenarios. The experiments show thatprefs2vec with a standard memory-based recommender system outperforms all the state-of-the-art baselines in terms of ranking accuracy, diversity, and novelty.