Research map / Large language models / Retrieval and knowledge
ConvMix: A Mixed-Criteria Data Augmentation Framework for Conversational Dense Retrieval
Fengran Mo, Jinghan Zhang, Yuchen Hui, Jia Ao Sun, Zhichao Xu, Zhan Su, Jian-Yun Nie
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 |
|---|---|---|---|
| add FIM-based utilization selection (ablation) | ConvMix-Combine · TopiOCQA | MRR: 33.0 -> 37.7 | Improved |
| add semantic diversity selection (k-means) (ablation) | ConvMix-Combine · TopiOCQA | MRR: 36.2 -> 37.7 | Improved |
| add query-side augmentation (ablation) | ConvMix-Combine · TopiOCQA | MRR: 34.4 -> 37.7 | Improved |
| add document-side augmentation (ablation) | ConvMix-Combine · CAsT-21 (out-of-domain) | MRR: 49.9 -> 60.3 | Improved |
| add document-side augmentation (ablation) | ConvMix-Combine · TopiOCQA | MRR: 36.9 -> 37.7 | No clear effect |
| include original data alongside augmented data (ablation) | ConvMix-Combine · CAsT-20 (out-of-domain) | MRR: 36.7 -> 46.1 | Improved |
| add topic-order shuffle augmentation to ANCE (ablation) | ANCE · TopiOCQA | MRR: 22.9 -> 29.8 | Improved |
| add topic-order shuffle augmentation to ANCE (ablation) | ANCE · QReCC | MRR: 46.3 -> 49.1 | Improved |
| scale augmented data from 1-fold to 10-fold (ablation) | ConvMix-Combine · TopiOCQA / QReCC | MRR | Mixed |
| ConvMix-Combine vs ConvAUG on QReCC | ConvMix-Combine · QReCC | Recall@10: 75.6 -> 75.6 | No clear effect |
| ConvMix-Combine vs ConvTrans on CAsT-19 | ConvMix-Combine · CAsT-19 (out-of-domain) | MRR: 73.2 -> 68.2 | Hurt |
Closest papers by meaning
- ChatQA: Surpassing GPT-4 on Conversational QA and RAG (NeurIPS 2024 · similarity 0.87)
- SeCom: On Memory Construction and Retrieval for Personalized Conversational Agents (ICLR 2025 · similarity 0.87)
- Cooperative Retrieval-Augmented Generation for Question Answering: Mutual Information Exchange and Ranking by Contrasting Layers (NeurIPS 2025 · similarity 0.86)
- ESPRESSO: An Effective Approach to Passage Retrieval for High-Quality Conversational Recommender Systems (AAAI 2025 · similarity 0.86)
- Chat-based Person Retrieval via Dialogue-Refined Cross-Modal Alignment (CVPR 2025 · similarity 0.86)
- Making Retrieval-Augmented Language Models Robust to Irrelevant Context (ICLR 2024 · 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.
