RecJourney 是一个专注于推荐系统算法实现与实验的开源项目。本项目旨在提供一个全面的推荐系统学习和研究平台,包含多个主流推荐算法的PyTorch实现,以及在公开数据集上的实验结果。
- 📚 包含20+种主流推荐算法的PyTorch实现
- 🔬 在多个公开数据集上进行了实验对比
- 🛠 提供了完整的数据预处理、模型训练、评估流程
- 📊 详细的实验结果分析与可视化
- 🔧 模块化设计,易于扩展新的算法
- Python 3.7+
- PyTorch 1.7+
- CUDA 10.1+ (对于GPU加速)
- 克隆仓库:
git clone https://github.com/yourusername/RecJourney.git
- 安装依赖:
pip install -r requirements.txt
TODO
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
Deep Interest Network for Click-Through Rate Prediction | KDD 2018 | http://arxiv.org/abs/1706.06978 | models/DIN |
Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate | SIGIR 2018 | https://arxiv.org/pdf/1804.07931 | models/ESMM |
Deep Interest Evolution Network for Click-Through Rate Prediction | AAAI 2019 | http://arxiv.org/abs/1809.03672 | models/DIEN |
Deep Session Interest Network for Click-Through Rate Prediction | IJCAI 2019 | http://arxiv.org/abs/1905.06482 | models/DSIN |
Behavior Sequence Transformer for E-commerce Recommendation in Alibaba | KDD 2019 | http://arxiv.org/abs/1905.06874 | models/BST |
Hybrid Contrastive Constraints for Multi-Scenario Ad Ranking | CIKM 2023 | https://arxiv.org/abs/2302.02636 | models/HC2 |
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
Deep & Cross Network for Ad Click Predictions | KDD 2017 | http://arxiv.org/abs/1708.05123 | models/DCN |
DCN V2: Improved Deep & Cross Network and Practical Lessons for Web-scale Learning to Rank Systems | WWW2021 | https://arxiv.org/pdf/2008.13535 | models/DCN |
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
Enhancing Explicit and Implicit Feature Interactions via Information Sharing for Parallel Deep CTR Models | KDD 2021 | https://dlp-kdd.github.io/assets/pdf/DLP-KDD_2021_paper_12.pdf | models/EDCN |
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
Progressive Layered Extraction (PLE): A Novel Multi-Task Learning (MTL) Model for Personalized Recommendations | RecSys 2020 | https://github.com/tangxyw/RecSysPapers/blob/927b56b90a0d7252114131ad08e516c0ad143106/Multi-Task/%5B2020%5D%5BTencent%5D%5BPLE%5D%20Progressive%20Layered%20Extraction%20(PLE)%20-%20A%20Novel%20Multi-Task%20Learning%20(MTL)%20Model%20for%20Personalized%20Recommendations.pdf | modes/PLE |
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
POSO: Personalized Cold Start Modules for Large-scale Recommender Systems | 2021 | https://arxiv.org/pdf/2108.04690 | models/POSO |
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
HiNet: Novel Multi-Scenario & Multi-Task Learning with Hierarchical Information Extraction | ICDE 2023 | https://arxiv.org/pdf/2303.06095 | model/HiNet |
论文标题 | 发表年份 | URL | 仓库 |
---|---|---|---|
Towards Deeper, Lighter and Interpretable Cross Network for CTR Prediction | CIKM 2023 | https://arxiv.org/pdf/2311.04635 | models/GDCN |
模型名称 | 标签 | 进度 |
---|---|---|
AFM | 完成 | |
AutoInt | 完成 | |
BST | 完成 | |
CAN | 完成 | |
DCN | 完成 | |
DIEN | 完成 | |
DIN | 完成 | |
DSIN | ||
EDCN | 完成 | |
ESMM | 完成 | |
FiBiNet | ||
FM | ||
GDCN | 完成 | |
HC2 | ||
HiNet | 完成 | |
MMOE | ||
PEPNet | ||
PLE | 完成 | |
POSO | 完成 | |
WideDeep | 完成 | |
xDeepFM | 完成 |
数据集:IJCAI 18
模型名称 | AUC | LogLoss |
---|---|---|
AutoInt | 0.56431 | 0.09450 |
SharedBottom | 0.55899 | 0.09820 |
TODO
TODO