Interest Clock: Time Perception in Real-Time Streaming Recommendation System

@article{Zhu2024InterestCT,
  title={Interest Clock: Time Perception in Real-Time Streaming Recommendation System},
  author={Yongchun Zhu and Jingwu Chen and Ling Chen and Yitan Li and Feng Zhang and Zuotao Liu},
  journal={ArXiv},
  year={2024},
  volume={abs/2404.19357},
  url={https://api.semanticscholar.org/CorpusID:269457305}
}
This paper proposes an effective and universal method Interest Clock to perceive time information in recommendation systems, which encodes users' time-aware preferences into a clock and then uses Gaussian distribution to smooth and aggregate them into the final interest clock embedding according to the current time for the final prediction.
[PDF] Semantic Reader

Figures and Tables from this paper

5 Citations

Long-Term Interest Clock: Fine-Grained Time Perception in Streaming Recommendation System

A fine-grained method for perceiving time information for streaming recommendation systems, named Long-term Interest Clock, which has been integrated into Douyin Music App's recommendation system and is capable of capturing users' dynamic fine-grained interests from long-term behaviors.

Moment&Cross: Next-Generation Real-Time Cross-Domain CTR Prediction for Live-Streaming Recommendation at Kuaishou

Kuaishou, is one of the largest short-video and live-streaming platform, and the amount of exposed short-video is 9x more than exposed live-streaming, which leads to a serious data imbalance problem making the live-streaming data could not fully reflect user interests.

AdaF^2M^2: Comprehensive Learning and Responsive Leveraging Features in Recommendation System

A model-agnostic framework AdaF^2M^2, short for Adaptive Feature Modeling with Feature Mask, which helps comprehensive feature learning via multi-forward training with augmented samples, while the adapter applies adaptive weights on features responsive to different user/item states.

CRM: Retrieval Model with Controllable Condition

The Controllable Retrieval Model (CRM), which integrates regression information as conditional features into the two-tower retrieval paradigm is proposed, which enables the retrieval stage could fulfill the target gap with ranking model, enhancing the retrieval model ability to search item candidates satisfied the user interests and condition effectively.

Multi-interest sequential recommendation with contrastive learning and temporal analysis

12 References

TWIN: TWo-stage Interest Network for Lifelong User Behavior Modeling in CTR Prediction at Kuaishou

This work proposes TWo-stage Interest Network (TWIN), where the Consistency-Preserved GSU (CP-GSU) adopts the identical target-behavior relevance metric as the TA in ESU, making the two stages twins.

Search-based User Interest Modeling with Lifelong Sequential Behavior Data for Click-Through Rate Prediction

A new cascaded search paradigm enables SIM with a better ability to model lifelong sequential behavior data in both scalability and accuracy, and also introduces the hands-on experience on how to implement SIM in large scale industrial systems.

Personalized Top-N Sequential Recommendation via Convolutional Sequence Embedding

A Convolutional Sequence Embedding Recommendation Model »Caser» 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.

User Consumption Intention Prediction in Meituan

In Meituan, a real-world system consisting of two stages, intention detection and prediction, a graph neural network-based intention prediction model GRIP is designed, which can capture user intrinsic preference and spatio-temporal context.

Deep Interest Network for Click-Through Rate Prediction

A novel model: Deep Interest Network (DIN) is proposed which tackles this challenge by designing a local activation unit to adaptively learn the representation of user interests from historical behaviors with respect to a certain ad.

Modeling Dual Period-Varying Preferences for Takeaway Recommendation

This work designs a dual interaction-aware module, aiming to capture users' dual preferences based on their interactions with stores and foods, and demonstrates that this model outperforms state-of-the-art methods on real-world datasets and is capable of modeling the dual period-varying preferences.

Collaborative Knowledge Base Embedding for Recommender Systems

A heterogeneous network embedding method is adopted, termed as TransR, to extract items' structural representations by considering the heterogeneity of both nodes and relationships and a final integrated framework, which is termed as Collaborative Knowledge Base Embedding (CKE), to jointly learn the latent representations in collaborative filtering.

Automatically Discovering User Consumption Intents in Meituan

This work designs the AutoIntent system, consisting of the disentangled intent encoder and intent discovery decoder, and proposes to build intent-pair pseudo labels based on the denoised feature similarities to transfer knowledge from known intents to new ones.

Deep Neural Networks for YouTube Recommendations

This paper details a deep candidate generation model and then describes a separate deep ranking model and provides practical lessons and insights derived from designing, iterating and maintaining a massive recommendation system with enormous user-facing impact.

DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems

This work proposes an improved framework DCN-V2, which is simple, can be easily adopted as building blocks, and has delivered significant offline accuracy and online business metrics gains across many web-scale learning to rank systems at Google.