Disentangled Item Representation for Recommender Systems

  title={Disentangled Item Representation for Recommender Systems},
  author={Zeyu Cui and Feng Yu and Shu Wu and Q. Liu and Liang Wang},
  journal={ACM Transactions on Intelligent Systems and Technology (TIST)},
  pages={1 - 20}
  • Zeyu Cui, Feng Yu, +2 authors Liang Wang
  • Published 17 August 2020
  • Computer Science
  • ACM Transactions on Intelligent Systems and Technology (TIST)
Item representations in recommendation systems are expected to reveal the properties of items. Collaborative recommender methods usually represent an item as one single latent vector. Nowadays the e-commercial platforms provide various kinds of attribute information for items (e.g., category, price, and style of clothing). Utilizing this attribute information for better item representations is popular in recent years. Some studies use the given attribute information as side information, which… 
Controllable Recommenders using Deep Generative Models and Disentanglement
It is demonstrated that a controllable recommender can be trained with a slight reduction in recommender performance, provided enough supervision is provided, and the recommendations produced appear to both conform to a user’s current preference and remain personalized.
Latent Structures Mining with Contrastive Modality Fusion for Multimedia Recommendation
In the proposed MICRO model, a novel modality-aware structure learning module is devised, which learns item-item relationships for each modality, and a novel multi-modal contrastive framework is designed to facilitate fine-grained multimodal fusion.
Mining Latent Structures for Multimedia Recommendation
A novel modality-aware structure learning layer is devised, which learns item-item structures for each modality and aggregates multiple modalities to obtain latent item graphs and performs graph convolutions to explicitly inject high-order item affinities into item representations.


VBPR: Visual Bayesian Personalized Ranking from Implicit Feedback
This paper proposes a scalable factorization model to incorporate visual signals into predictors of people's opinions, which is applied to a selection of large, real-world datasets and makes use of visual features extracted from product images using (pre-trained) deep networks.
MRLR: Multi-level Representation Learning for Personalized Ranking in Recommendation
A unified Bayesian framework MRLR is designed to learn user and item embeddings from a multi-level item organization, thus benefiting from RL as well as achieving the goal of personalized ranking.
Differentiated Fashion Recommendation Using Knowledge Graph and Data Augmentation
A differentiated recommendation framework is proposed that provides different recommendation paths for active and inactive users to improve the overall recommendation quality and the experimental results show that through data augmentation algorithm to improve data quality, factorization machine model produces higher recommendation accuracy.
BPR: Bayesian Personalized Ranking from Implicit Feedback
This paper presents a generic optimization criterion BPR-Opt for personalized ranking that is the maximum posterior estimator derived from a Bayesian analysis of the problem and provides a generic learning algorithm for optimizing models with respect to B PR-Opt.
Explainable Fashion Recommendation: A Semantic Attribute Region Guided Approach
This work proposes a novel Semantic Attribute Explainable Recommender System (SAERS), capable of not only providing cloth recommendations for users, but also explaining the reason why the authors recommend the cloth through intuitive visual attribute semantic highlights in a personalized manner.
DeepStyle: Learning User Preferences for Visual Recommendation
A DeepStyle method is proposed for learning style features of items and sensing preferences of users and the effectiveness of DeepStyle for visual recommendation is illustrated.
Sherlock: Sparse Hierarchical Embeddings for Visually-Aware One-Class Collaborative Filtering
A novel hierarchical embedding architecture is built to model the visual dimensions across different product categories with both high-level (colorfulness, darkness, etc.) and subtle (e.g. casualness) visual characteristics simultaneously.
Factorizing personalized Markov chains for next-basket recommendation
This paper introduces an adaption of the Bayesian Personalized Ranking (BPR) framework for sequential basket data and shows that the FPMC model outperforms both the common matrix factorization and the unpersonalized MC model both learned with and without factorization.
Collaborative Preference Embedding against Sparse Labels
This work proposes a novel method named as Collaborative Preference Embedding (CPE) which directly deals with sparse and insufficient user preference information and leverages a compact embedding space by reducing the dependency across different dimensions of a code (embedding).
Transfer Meets Hybrid: A Synthetic Approach for Cross-Domain Collaborative Filtering with Text
The proposed Transfer Meeting Hybrid (TMH) model attentively extracts useful content from unstructured text via a memory network and selectively transfers knowledge from a source domain via a transfer network and shows better performance in terms of three ranking metrics by comparing with various baselines.