Research map / Multi-agent systems and games / Human-AI and goal-conditioned
Keyword-Guided Neural Conversational Model
Peixiang Zhong, Yong Liu, Hao Wang, Chunyan Miao
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 CKG-aware concept representations (ablation) | CKC next-turn keyword prediction · ConvAI2 test | R@1: 18.56±0.31 -> 19.31±0.44 | Improved |
| add CKG-aware concept representations (ablation) | CKC keyword-augmented response retrieval · ConvAI2 test | R@1: 53.11±0.43 -> 59.90±0.41 | Improved |
| add separate keyword matching (ablation) | CKC keyword-augmented response retrieval · ConvAI2 test | R@1: 52.30±0.54 -> 59.90±0.41 | Improved |
| use CKG-guided keyword selection instead of embedding-based selection (ablation) | CKC self-play · ConvAI2 | target success rate: 22.3 -> 28.9 | Improved |
| use CKG-guided keyword selection instead of embedding-based selection (ablation) | CKC self-play · ConvAI2 | turns to reach target: 4.42 -> 4.23 | Improved |
| incorporate CKG triplets into next-turn keyword prediction | CKC · ConvAI2 and Reddit | R@1: 18.03±0.15 -> 19.31±0.44 (ConvAI2); 7.11±0.21 -> 8.23±0.31 (Reddit) | Improved |
| use CKC for keyword-augmented response retrieval | keyword-augmented response retrieval · ConvAI2 and Reddit | R@1: 48.67±0.25 -> 59.90±0.41 (ConvAI2); 45.31±0.70 -> 50.02±0.41 (Reddit) | Improved |
| use CKC for keyword-guided dialogue | self-play · ConvAI2 and Reddit | target success rate: 25.6 -> 28.9 (ConvAI2); 18.4 -> 22.7 (Reddit) | Improved |
| use CKC for keyword-guided dialogue | self-play · ConvAI2 and Reddit | turns to reach target: 4.54 -> 4.23 (ConvAI2); 4.42 -> 4.19 (Reddit) | Improved |
| use CKC for keyword-guided dialogue | human-evaluated conversations · ConvAI2 and Reddit | conversation smoothness: 3.01 -> 3.27 (ConvAI2); 2.81 -> 3.08 (Reddit) | Improved |
Closest papers by meaning
- Learning to Clarify: Multi-turn Conversations with Action-Based Contrastive Self-Training (ICLR 2025 · similarity 0.86)
- SeCom: On Memory Construction and Retrieval for Personalized Conversational Agents (ICLR 2025 · similarity 0.85)
- Beyond Prompts: Dynamic Conversational Benchmarking of Large Language Models (NeurIPS 2024 · similarity 0.85)
- Artificial Hivemind: The Open-Ended Homogeneity of Language Models (and Beyond) (NeurIPS 2025 · similarity 0.84)
- The Consensus Game: Language Model Generation via Equilibrium Search (ICLR 2024 · similarity 0.84)
- A Statistical Framework for Ranking LLM-based Chatbots (ICLR 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.
