Research map / Probabilistic and scientific modeling / Treatment effects
Estimating Propensity for Causality-based Recommendation without Exposure Data
Zhongzhou Liu, Yuan Fang, Min Wu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Estimate propensity and exposure from interactions using pairwise popularity prior and beta regularization | DLCE recommendation pipeline · DH_original, DH_personalized, and ML | CP@10, CP@100, CDCG: .0351 vs .0263 CP@10; .1270 vs .0564 CP@10; .0182 vs .0124 CP@10 | Improved |
| Remove the propensity constraint from the pairwise popularity loss (NO_P) (ablation) | PropCare · Causal recommendation | causal performance | Hurt |
| Remove the relevance constraint from the pairwise popularity loss (NO_R) (ablation) | PropCare · Causal recommendation | causal performance | Hurt |
| Remove the entire pairwise popularity loss (NO_P_R) (ablation) | PropCare · Causal recommendation | causal performance | Hurt |
| Reverse the assumption so more popular items have smaller propensity scores (NEG) (ablation) | PropCare · Causal recommendation | causal performance | Hurt |
| Set all pairwise weights κ to 1 (ablation) | PropCare · Causal recommendation | causal performance | Hurt |
| Add beta-distribution propensity regularization instead of removing it (μ=0) (ablation) | PropCare · Causal recommendation | causal performance | Improved |
| Increase propensity or exposure estimation error | DLCE · Causal recommendation | causal performance | Hurt |
| Use interaction-based ranking that ignores exposure status | MF, BPR, and LightGCN · DH_original, DH_personalized, and ML | causal performance | Hurt |
Closest papers by meaning
- Causal Representation Learning for Generalisable Recommendation (NeurIPS 2026 · similarity 0.88)
- Transferring Causal Effects using Proxies (NeurIPS 2025 · similarity 0.85)
- Alleviating Dual Biases in Recommendation (Student Abstract) (AAAI 2025 · similarity 0.85)
- Debiased Collaborative Filtering with Kernel-Based Causal Balancing (ICLR 2024 · similarity 0.85)
- Addressing Hidden Confounding with Heterogeneous Observational Datasets for Recommendation (NeurIPS 2024 · similarity 0.84)
- Telling Peer Direct Effects from Indirect Effects in Observational Network Data (ICML 2025 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
