Research map / Label-efficient and robust learning / Recognition and long-tail
A Speaker Turn-Aware Multi-Task Adversarial Network for Joint User Satisfaction Estimation and Sentiment Analysis
Kaisong Song, Yangyang Kang, Jiawei Liu, Xurui Li, Changlong Sun, Xiaozhong Liu
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 |
|---|---|---|---|
| Use STMAN's task-specific adversarial learning and speaker-turn-aware interaction | joint USE and SA · Clothes | USE Macro F1: 70.40 -> 71.11 | Improved |
| Use STMAN's task-specific adversarial learning and speaker-turn-aware interaction | joint USE and SA · Makeup | USE Macro F1: 79.18 -> 80.11 | Improved |
| Use STMAN's task-specific adversarial learning and speaker-turn-aware interaction | joint USE and SA · Clothes | SA Macro F1: 80.29 -> 81.90 | Improved |
| Use STMAN's task-specific adversarial learning and speaker-turn-aware interaction | joint USE and SA · Makeup | SA Macro F1: 83.36 -> 84.18 | Improved |
| Add a role-selected mask to focus the satisfaction representation on user utterances (ablation) | Basic USE model · Clothes | USE Macro F1: 67.29 -> 67.67 | Improved |
| Add a role-selected mask to focus the satisfaction representation on user utterances (ablation) | Basic USE model · Makeup | USE Macro F1: 77.12 -> 78.69 | Improved |
| Add the auxiliary sentiment analysis task (ablation) | Basic USE model · Clothes | USE Macro F1: 67.03 -> 67.67 | Improved |
| Add the auxiliary sentiment analysis task (ablation) | Basic USE model · Makeup | USE Macro F1: 74.64 -> 78.69 | Improved |
| Add a task discriminator to separate task-specific features (ablation) | Basic multi-task model · Clothes | USE Macro F1: 67.67 -> 68.92 | Improved |
| Add speaker-turn change information (ablation) | Basic multi-task model · Clothes | USE Macro F1: 67.67 -> 69.53 | Improved |
| Add a task discriminator to separate task-specific features (ablation) | Basic multi-task model · Makeup | USE Macro F1: 78.69 -> 79.05 | Improved |
| Add speaker-turn change information (ablation) | Basic multi-task model · Makeup | USE Macro F1: 78.69 -> 79.03 | Improved |
Closest papers by meaning
- Re-Evaluating ADEM: A Deeper Look at Scoring Dialogue Responses (AAAI 2019 · similarity 0.86)
- Filling the Gap of Utterance-aware and Speaker-aware Representation for Multi-turn Dialogue (AAAI 2021 · similarity 0.84)
- A Comprehensive Analysis of the Effectiveness of Large Language Models as Automatic Dialogue Evaluators (AAAI 2024 · similarity 0.84)
- Aligning Spoken Dialogue Models from User Interactions (ICML 2025 · similarity 0.82)
- An Efficient Self-Learning Framework For Interactive Spoken Dialog Systems (ICML 2024 · similarity 0.81)
- Dynamic-SUPERB Phase-2: A Collaboratively Expanding Benchmark for Measuring the Capabilities of Spoken Language Models with 180 Tasks (ICLR 2025 · similarity 0.81)
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.
