Research map / Learning theory and algorithms / Prediction and decision losses
Rank Aggregation via Heterogeneous Thurstone Preference Models
Tao Jin, Pan Xu, Quanquan Gu, Farzad Farnoud
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 |
|---|---|---|---|
| Model each worker's accuracy level instead of using a uniform-accuracy BTL model | HBTL rank aggregation · Reading Level | Kendall's tau: 0.3472 -> 0.3763 | Improved |
| Model each worker's accuracy level instead of using a uniform-accuracy BTL model | HBTL rank aggregation · Country Population | Kendall's tau: 0.7524 -> 0.7905 | Improved |
| Model heterogeneous user accuracy with the normal-noise HTCV model rather than uniform TCV | HTCV rank aggregation · Reading Level | Kendall's tau: 0.3452 -> 0.3729 | Improved |
| Model heterogeneous user accuracy with the normal-noise HTCV model rather than uniform TCV | HTCV rank aggregation · Country Population | Kendall's tau: 0.7524 -> 0.7714 | Improved |
| Use HBTL rather than CrowdBT | Reading Level rank aggregation · Reading Level | Kendall's tau: 0.3737 -> 0.3763 | Improved |
| Use HBTL rather than CrowdBT | Country Population rank aggregation · Country Population | Kendall's tau: 0.7714 -> 0.7905 | Improved |
| Use HTCV rather than CrowdTCV | Reading Level rank aggregation · Reading Level | Kendall's tau: 0.3672 -> 0.3729 | Improved |
| Use HTCV rather than CrowdTCV | Country Population rank aggregation · Country Population | Kendall's tau: 0.7714 -> 0.7714 | No clear effect |
| Set HBTL virtual-node regularization λ₀ from 0 to 10 (ablation) | HBTL · Reading Level | Kendall's tau: 0.3608 -> 0.3763 | Improved |
| Set HBTL virtual-node regularization λ₀ from 0 to 5 (ablation) | HBTL · Country Population | Kendall's tau: 0.7905 -> 0.7524 | Hurt |
Closest papers by meaning
- Data-driven Rank Breaking for Efficient Rank Aggregation (ICML 2016 · similarity 0.87)
- Optimal Sample Complexity of M-wise Data for Top-K Ranking (NeurIPS 2017 · similarity 0.87)
- Accelerated Spectral Ranking (ICML 2018 · similarity 0.87)
- Statistical Inference for Incomplete Ranking Data: The Case of Rank-Dependent Coarsening (ICML 2017 · similarity 0.87)
- Aggregating Quantitative Relative Judgments: From Social Choice to Ranking Prediction (NeurIPS 2024 · similarity 0.86)
- Selective Preference Aggregation (ICML 2025 · similarity 0.85)
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.
