Research map / Label-efficient and robust learning / Recognition and long-tail
Unsupervised Summarization for Chat Logs with Topic-Oriented Ranking and Context-Aware Auto-Encoders
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 context-aware DAE compression to the topic utterance extractor (ablation) | RankAE · E-commerce chat log test set | ROUGE-1: +2.53 | Helped |
| Add context-aware DAE compression to the topic utterance extractor (ablation) | RankAE · E-commerce chat log test set | ROUGE-2: +1.72 | Helped |
| Add context-aware DAE compression to the topic utterance extractor (ablation) | RankAE · E-commerce chat log test set | ROUGE-L: +3.22 | Helped |
| Add the utterance-distance constraint to topic ranking (ablation) | RankAE (Ext.) · E-commerce chat log test set | ROUGE-1: 25.68 -> 27.19 | Helped |
| Add diversity enhancement to topic ranking (ablation) | RankAE (Ext.) · E-commerce chat log test set | ROUGE-1: 24.11 -> 27.19 | Helped |
| Add context utterances with window size c = 1 (ablation) | RankAE (Ext.) · E-commerce chat log test set | ROUGE-1: 27.19 -> 34.66 | Helped |
| Increase context window size from c = 1 to c = 2 (ablation) | RankAE (Ext.) + context · E-commerce chat log test set | BLEU: 16.92 -> 16.75 | Hurt |
| Add noise to DAE training at the reported rates (ablation) | RankAE · E-commerce chat log test set | ROUGE-1: 34.53 -> 32.34; 30.49; 30.30 | Hurt |
| Add noise to DAE training at the reported rates (ablation) | RankAE · E-commerce chat log test set | BLEU: 16.90 -> 17.28; 17.62; 17.85 | Helped |
| Use BERT-pretrained utterance representations rather than the un-pretrained Transformer encoder | RankAE · E-commerce chat log test set | ROUGE-1: 27.63 -> 28.20 | Helped |
| Include surrounding context without DAE compression (ablation) | RankAE (Ext.) · E-commerce chat log test set | BLEU: 17.31 -> 16.92 | Hurt |
Closest papers by meaning
- Unsupervised Abstractive Dialogue Summarization for Tete-a-Tetes (AAAI 2021 · similarity 0.87)
- DialogLM: Pre-trained Model for Long Dialogue Understanding and Summarization (AAAI 2022 · similarity 0.86)
- Unsupervised Extractive Summarization with Learnable Length Control Strategies (AAAI 2024 · similarity 0.84)
- PersonalSum: A User-Subjective Guided Personalized Summarization Dataset for Large Language Models (NeurIPS 2024 · similarity 0.83)
- Learning from Failure: Improving Meeting Summarization without Good Samples (AAAI 2024 · similarity 0.82)
- Document Summarization with Conformal Importance Guarantees (NeurIPS 2025 · similarity 0.82)
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.