Latent Relational Metric Learning via Memory-based Attention for Collaborative Ranking

By Yi Tay, Anh Tuan Luu, Siu Cheung Hui. The Web Conference, 2017

322 citations recorded by the source corpus.

This paper proposes a new neural architecture for collaborative ranking with implicit feedback. Our model, LRML (Latent Relational Metric Learning) is a novel metric learning approach for recommendation. More specifically, instead of simple push-pull mechanisms between user and item pairs, we propose to learn latent relations that describe each user item interaction. This helps to alleviate the potential geometric inflexibility of existing metric learning approaches. This enables not only better performance but also a greater extent of modeling capability, allowing our model to scale to a larger number of interactions. In order to do so, we employ a augmented memory module and learn to attend over these memory blocks to construct latent relations. The memory-based attention module is controlled by the user-item interaction, making the learned relation vector specific to each user-item pair. Hence, this can be interpreted as learning an exclusive and optimal relational translation for each user-item interaction. The proposed architecture demonstrates the state-of-the-art performance across multiple recommendation benchmarks. LRML outperforms other metric learning models by 6%-7.5% in terms of Hits@10 and nDCG@10 on large datasets such as Netflix and MovieLens20M. Moreover, qualitative studies also demonstrate evidence that our proposed model is able to infer and encode explicit sentiment, temporal and attribute information despite being only trained on implicit feedback. As such, this ascertains the ability of LRML to uncover hidden relational structure within implicit datasets.

Subfields: Artificial Intelligence, Information Systems

View Paper: https://www.semanticscholar.org/paper/23b0a959d07a744a80734119e94262f14f2d83e1

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Xiangnan He
2. Xiang Wang
3. Tat-seng Chua
4. Fuli Feng
5. Yi Tay
6. Meng Wang
7. Chuan Shi
8. Qiang Yang

In Information Systems

1. Xiangnan He
2. Xiang Wang
3. Tat-seng Chua
4. Meng Wang
5. Fuli Feng
6. Yi Tay
7. Naomie Binti Salim
8. Aminu Da’u

Citations in the corpus

• Unifying Knowledge Graph Learning and Recommendation: Towards a Better Understanding of User Preferences (2019, background): “h , e ′ t , r ′ )∈KG − [ f ( e h , e t , r ) + γ − f ( e ′ h , e ′ t , r ′ )] + (4) where [·] + ≜ max ( 0 , ·) , KG − contains incorrect triplets constructed by replacing head entity or tail entity in a valid triplet randomly, and γ controls the margin between positive and negative triplets.”
• Transfer Meets Hybrid: A Synthetic Approach for Cross-Domain Collaborative Filtering with Text (2019, methodology): “Memory networks have been used in recommendation to model item content [22], model users’ neighborhood [12], and learn latent relationships [48].”
• Are we really making much progress? A worrying analysis of recent neural recommendation approaches (2019, unknown): “Non-reproducible: KDD: [43], RecSys: [41], [6], [38], [44], [21], [45], SIGIR: [32], [7], WWW: [42], [11]”
• CLEAR: Contrastive-Prototype Learning with Drift Estimation for Resource Constrained Stream Mining (2021, background): “Contrastive Learning, also named as embedding learning, aims to learn a suitable embedding space [16, 24, 33, 40] to minimize the intra-class variation and maximize the inter-class variation.”
• Recommendation system based on deep learning methods: a systematic review and new directions (2019, methodology): “These show the promising potentials of the deep learning methods for the multitasking RS. Neural Attention Neural attention or attention mechanism is a technique that enables a neural network to focus on a subset of features by selecting a specific input (Tay et al. 2018b).”
• NAIS: Neural Attentive Item Similarity Model for Recommendation (2018, methodology): “[41] develops an attentive memory network for learning the user-item scoring function, and Wang et al.”
• Disentangled Graph Collaborative Filtering (2020, background): “Many recommenders, such as NCF [12], CMN [7], and LRML [35] resort to this paradigm.”
• LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation (2020, methodology): “The recent neural recommender models like NCF [19] and LRML [34] use the same embedding component, while enhance the interaction modeling with neural networks.”
• Deep Learning Based Recommender System (2017, methodology): “Attention-based CNNs are capable of capturing the most informative elements of the inputs [128].”
• Where to Go Next: Modeling Long- and Short-Term User Preferences for Point-of-Interest Recommendation (2020, background): “…such as embedding learning (Feng et al. 2015; Shi et al. 2018; Zhang et al. 2019), neural CF (He et al. 2017; Yin et al. 2017), deep latent factor model (Cheng et al. 2018), and metric learning (Tay, Anh Tuan, and Hui 2018) models, achieve promising performance in many recommendation systems.”
• A Survey on Accuracy-Oriented Neural Recommendation: From Collaborative Filtering to Information-Rich Recommendation (2021, background): “Distinct from CML that uses simple metric learning that assumes each user’s embedding is equally close to every item embedding she likes, LRML introduces the relation vectors r to capture the relationships between user and item pairs [47] .”
• Neural Graph Collaborative Filtering (2019, methodology): “For instance, neural CF models, such as NeuMF [14], employ nonlinear neural networks as the interaction function; meanwhile, translation-based CF models, such as LRML [28], instead model the interaction strength with Euclidean distance metrics.”
• Cross-domain Recommendation Without Sharing User-relevant Data (2019, unknown): “Tay et al. [36] proposed a relational-translation based match function to learn from interactions.”
• Multi-Pointer Co-Attention Networks for Recommendation (2018, background): “Today, we see a shift into deep learning, in which neural models are claiming state-of-the-art performance [10, 27, 36, 38].”
• Leveraging Meta-path based Context for Top- N Recommendation with A Neural Co-Attention Model (2018, background): “For the other two datasets, we follow [11, 32] to treat a rating as an interaction record, indicating whether a user has rated an item.”