目录结构：
- code/ 源码（官方）
-
note/ 笔记
-
paper/ 论文原文
论文列表：
| 索引 | 标题 | 简介 | 代码 |
|---|---|---|---|
| 1 | Improving Ad Click Prediction by Considering Non-displayed Events | 利用无偏集合S_{t}去偏，IPS / CausE / New + FFM | https://www.csie.ntu.edu.tw/~cjlin/papers/occtr/ / https://github.com/jyhsia5174/CIKM-2019-EXP (c++) |
| 2 | Recommendations as Treatments: Debiasing Learning and Evaluation | IPS Estimator | https://www.cs.cornell.edu/~schnabts/mnar/ (python) |
| 3 | Causal Embeddings for Recommendation | Cause Embedding \pi_{c},\pi_{t}, 假设有 S_{c}, S_{t} 一起计算两个矩阵 | https://github.com/criteo-research/CausE 缺少数据处理部分(tensorflow) |
| 4 | Causal Inference for Recommendation | PS / 重构测试集(使得训练集和测试集不同分布)比较因果方法与传统方法 | https://github.com/gizemt/CausalRecommendation (.m?) |
| 5 | Bias and Debias in Recommender System: A Survey and Future Directions | survey | |
| 6 | The Deconfounded Recommender: A Causal Inference Approach to Recommendation | PF -> exposure model / MF -> outcome model | |
| 7 | Disentangling User Interest and Popularity Bias for Recommendation with Causal Embedding | ||
| 8 | Debiasing Item-to-Item Recommendations With Small Annotated Datasets | ||
| 9 | Unbiased Ad Click Prediction for Position-aware Advertising | https://www.csie.ntu.edu.tw/~cjlin/papers/debiases/ / abstract / introduction | |
| 10 | Unbiased Implicit Recommendation and Propensity Estimation via Combinational Joint Learning | abstract / introduction | |
| 11 | Unbiased Learning for the Causal Effect of Recommendation | IPS CIPS / ERM 重点看 | |
| 12 | Unbiased Recommender Learning from Missing-Not-At-Random Implicit Feedback | abstract / introduction | |
| 13 | Collaborative prediction and ranking with non-random missing data | Propensity Estimation via Naive Bayes —— from [2] | |
| 14 | Observational Studies | Propensity Estimation via Logistic Regression —— from [2] | |
| 15 | Efficient estimation of average treatment effects using the estimated propensity score | IPS —— from [11] |