arXiv cs.CL· Y. L. Li, Y. H. Wu·· 11 小时前AI 评分32
有界递归自我改进框架提升长程角色扮演对话质量
Recursive Self-Improvement for Long-Horizon Role-Playing Dialogue
AI 导读
研究者提出一种有界递归自我改进（RSI）框架，通过修订角色记忆、稳定人设指令和构建逐轮提示词的策略来改进长程角色扮演对话。内部评测覆盖 50 个原创角色与 5 个模型、每对 2 段 100 轮对话，汇总回复的坏回复率从 14.00%（6,973/49,824）降至 4.94%（2,459/49,793），相对下降 64.71%。
正文 · 原文
Abstract:Long-horizon role-playing dialogue requires consistent character identity, reliable memory, and context-sensitive interaction. We present a bounded recursive self-improvement (RSI) framework that revises original-character memory, stable persona instructions, and a policy for constructing turn-level prompts. Repair outcomes also inform revisions to the updater's diagnostic rules, edit routing, and candidate selection. We specify the updater as a versioned policy with explicit inputs, editable fields, acceptance gates, and stopping conditions. An internal evaluation covers 50 original characters and five models, with two 100-round conversations per character-model pair: 500 conversations per condition. Across the pooled evaluated replies, the bad-response rate decreases from 14.00% (6,973/49,824) to 4.94% (2,459/49,793), a 9.06-percentage-point reduction and a 64.71% relative reduction. In 600 blinded human comparisons, RSI receives 294 wins, 187 ties, and 119 losses, yielding a 71.19% win rate among decisive judgments. The results support improved aggregate dialogue quality for the complete framework on the internal test distribution.
| Comments: | 8 pages, 1 figure, 4 tables |
| Subjects: | Computation and Language (cs.CL) |
| Cite as: | arXiv:2611.07469 [cs.CL] |
| (or arXiv:2611.07469v1 [cs.CL] for this version) | |
| https://doi.org/10.48550/arXiv.2611.07469 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yanlong Li [view email]
[v1]
Thu, 5 Nov 2026 05:28:10 UTC (308 KB)
来源：arXiv cs.CL · arxiv.org