Multi-Sparse-Domain Collaborative Recommendation via Enhanced Comprehensive Aspect Preference Learning

Abstract

Cross-domain recommendation (CDR) has been attracting increasing attention of researchers for its ability to alleviate the data sparsity problem in recommender systems. However, the existing singletarget or dual-target CDR methods often suffer from two drawbacks, the assumption of at least one rich domain and the heavy dependence on domain-invariant preference, which are impractical in real world where sparsity is ubiquitous and might degrade the user preference learning. To overcome these issues, we propose a Multi-Sparse-Domain Collaborative Recommendation (MSDCR) model for multi-target cross-domain recommendation. Unlike traditional CDR methods, MSDCR treats the multiple relevant domains as all sparse and can simultaneously improve the recommendation performance in each domain. We propose a Multi-Domain Separation Network (MDSN) and a Gated Aspect Preference Enhancement (GAPE) module for MSDCR to enhance a user's domain-specific aspect preferences in a domain by transferring the complementary aspect preferences in other domains, during which the uniqueness of the domain-specific preference can be preserved through the adversarial training offered by MDSN and the complementarity can be adaptively determined by GAPE. Meanwhile, we propose a Multi-Domain Adaptation Network (MDAN) for MSDCR to capture a user's domain-invariant aspect preference. With the integration of the enhanced domain-specific aspect preference and the domaininvariant aspect preference, MSDCR can reach a comprehensive understanding of a user's preference in each sparse domain. At last, the extensive experiments conducted on real datasets demonstrate the remarkable superiority of MSDCR over the state-of-the-art single-domain recommendation models and CDR models. CCS CONCEPTS• Information systems → Recommender systems.

PEACE: Prototype lEarning Augmented transferable framework for Cross-domain rEcommendation

“…As a promising direction, cross-domain recommendation (CDR) has attracted a surge of investigations, which enables the effective learning of a data-sparser domain by transferring useful knowledge from data-richer domains. Existing CDR methods often assume the existence of shared information so that a mapping function can be learned across different domains [27,31,39,58] or availability of source and target domains for joint optimization [20,23,56]. However, this assumption may not hold in real-world applications due to the considerable gap between source and target domains or even unavailability of target domains data during training, which severely hinders the application of existing CDR approaches.…”

Development of a New Reverse Genetics System for Ebola Virus

“…One possible and plausible recipe is cross-domain recommendation (CDR) [9,10], which utilize the behaviors of data-richer source domains (e.g., global behaviors that accumulate in online service platforms) for learning transferable knowledge to help improve recommendation in data-sparser target domains (e.g., miniapps). Existing CDR methods often assume that the availability of both source and target domains for learning mapping function across domains given the shared information [4,5] or for joint optimization [6,7]. However, in real-world service platforms, usually source and target domains can be highly diversified and sometimes the target domains are even not available during training, which influences the validity of this assumption and further affects the application of existing CDR approaches.…”