Continuous-Time Sequential Recommendation with Temporal Graph Collaborative Transformer

@article{Fan2021ContinuousTimeSR,
  title={Continuous-Time Sequential Recommendation with Temporal Graph Collaborative Transformer},
  author={Ziwei Fan and Zhiwei Liu and Jiawei Zhang and Yun Xiong and Lei Zheng and Philip S. Yu},
  journal={Proceedings of the 30th ACM International Conference on Information \& Knowledge Management},
  year={2021}
}
  • Ziwei Fan, Zhiwei Liu, Philip S. Yu
  • Published 14 August 2021
  • Computer Science
  • Proceedings of the 30th ACM International Conference on Information & Knowledge Management
In order to model the evolution of user preference, we should learn user/item embeddings based on time-ordered item purchasing sequences, which is defined as Sequential Recommendation~(SR) problem. Existing methods leverage sequential patterns to model item transitions. However, most of them ignore crucial temporal collaborative signals, which are latent in evolving user-item interactions and coexist with sequential patterns. Therefore, we propose to unify sequential patterns and temporal… 

Figures and Tables from this paper

Evolutionary Preference Learning via Graph Nested GRU ODE for Session-based Recommendation

Graph Nested GRU ordinary differential equation (ODE) is proposed, namely GNG-ODE, a novel continuum model that extends the idea of neural ODEs to continuous-time temporal session graphs and preserves the continuous nature of dynamic user preferences.

Time Interval-enhanced Graph Neural Network for Shared-account Cross-domain Sequential Recommendation

To fully account for users' domain-specific preferences on items, two effective attention mechanisms are further developed to selectively guide the message-passing process and incorporate the time interval into the message passing and design an account-aware self-attention module for learning items' interactive characteristics.

Sequential Recommendation via Stochastic Self-Attention

A novel Wasserstein Self-Attention module is devised to characterize item-item position-wise relationships in sequences, which effectively incorporates uncertainty into model training.

Long Short-Term Preference Modeling for Continuous-Time Sequential Recommendation

Long Short-Term Preference Modeling for Continuous-Time Sequential Recommendation (LSTSR) is proposed to capture the evolu- tion of short-term preference under dynamic graph.

Exploiting Session Information in BERT-based Session-aware Sequential Recommendation

Three methods to improve recommendation performance by exploiting session information while minimizing additional parameters in a BERT-based sequential recommendation model are proposed: using session tokens, adding session segment embeddings, and a time-aware self-attention.

Modeling Sequences as Distributions with Uncertainty for Sequential Recommendation

This work proposes a Distribution-based Transformer for Sequential Recommendation (DT4SR), which injects uncertainties into sequential modeling, and devise two novel Transformers for modeling mean and covariance, which guarantees the positive-definite property of distributions.

Sequential recommendation method integrating item relationship and user preference

The experimental results on two public datasets Beauty and Home show that the proposed method IRUP can effectively improve the recommendation performance, and the two evaluation metrics HR@5 and NDCG@5 have an average improvement of 8.28% and 7.03%, respectively.

Leaping Through Time with Gradient-based Adaptation for Recommendation

This work proposes a new solution named LeapRec to the temporal dynamic problem by using trajectory-based meta-learning to model time dependencies, and shows that LeapRec consistently outperforms the state-of-the-art methods on several datasets and recommendation metrics.

When Multi-Level Meets Multi-Interest: A Multi-Grained Neural Model for Sequential Recommendation

A unified multi-grained neural model (named MGNM) is proposed, which first learns the graph structure and information aggregation paths of the historical items for a user, then performs graph convolution to derive item representations in an iterative fashion, leading to a more precise interest learning in a multi-Grained manner.

Predicting Dynamic User-Item Interaction with Meta-Path Guided Recursive RNN

A Metapath-guided Recursive RNN based Shift embedding method named MRRNN-S is proposed to learn the continuously evolving embeddings of users and items for more accurately predicting their future interactions and achieves superior performance by extensive comparison with state-of-the-art baseline models.

References

SHOWING 1-10 OF 56 REFERENCES

Time Matters: Sequential Recommendation with Complex Temporal Information

This paper discovers two elementary temporal patterns of user behaviors and devise a neural architecture that jointly learns those temporal patterns to model user dynamic preferences and demonstrates the superiority of the model compared with the state-of-the-arts.

Time Interval Aware Self-Attention for Sequential Recommendation

This paper proposes TiSASRec (Time Interval aware Self-attention based sequential recommendation), which models both the absolute positions of items as well as the time intervals between them in a sequence, which outperforms various state-of-the-art sequential models on both sparse and dense datasets and different evaluation metrics.

Augmenting Sequential Recommendation with Pseudo-Prior Items via Reversely Pre-training Transformer

A new framework for Augmenting Sequential Recommendation with Pseudo-prior items (ASReP) is introduced, which firstly pre-train a transformer with sequences in a reverse direction to predict prior items, and uses this transformer to generate fabricated historical items at the beginning of short sequences.

DΓ©jΓ  vu: A Contextualized Temporal Attention Mechanism for Sequential Recommendation

This paper proposes a Contextualized Temporal Attention Mechanism that learns to weigh historical actions’ influence on not only what action it is, but also when and how the action took place and consistently outperformed an extensive set of state-of-the-art sequential recommendation methods.

Session-based Recommendation with Graph Neural Networks

In the proposed method, session sequences are modeled as graph-structured data and GNN can capture complex transitions of items, which are difficult to be revealed by previous conventional sequential methods.

Modeling Sequences as Distributions with Uncertainty for Sequential Recommendation

This work proposes a Distribution-based Transformer for Sequential Recommendation (DT4SR), which injects uncertainties into sequential modeling, and devise two novel Transformers for modeling mean and covariance, which guarantees the positive-definite property of distributions.

Temporal recommendation on graphs via long- and short-term preference fusion

This work proposes Session-based Temporal Graph (STG) which simultaneously models users' long-term and short-term preferences over time and proposes a novel recommendation algorithm Injected Preference Fusion (IPF) and extends the personalized Random Walk for temporal recommendation.

BasConv: Aggregating Heterogeneous Interactions for Basket Recommendation with Graph Convolutional Neural Network

A new framework named BasConv, which is based on the graph convolutional neural network, and has three types of aggregators specifically designed forThree types of nodes, which collectively learn node embeddings from both neighborhood and high-order context.

Personalized Top-N Sequential Recommendation via Convolutional Sequence Embedding

A Convolutional Sequence Embedding Recommendation Model Β»CaserΒ» is proposed, which is to embed a sequence of recent items into an image in the time and latent spaces and learn sequential patterns as local features of the image using convolutional filters.

Gated Spectral Units: Modeling Co-evolving Patterns for Sequential Recommendation

An efficient spectral convolution operation is introduced to discover complex relationships between users and items from the spectral domain of a graph, where the hidden connectivity information of the graph can be revealed and the advantages of modeling co-evolving patterns are demonstrated.
...