Journal of Digital Humanities and Science Fiction Studies
DHSFS, Vol. 1, No. 1, 2026, pp. 66-79.
Print ISSN: 3105-1278; Online ISSN: 3105-1286
Journal homepage: https://www.dhsfs.com
DOI: Https://doi.org/10.64058/DHSFS.26.1.06
数字文心:意义共振驱动的大模型叙述范式
徐晓雪(Xu Xiaoxue),范力欣(Fan Lixin)
摘要:为解决当前AI文学创作中的“无心”困境,本文创新性地提出了“数字文心”方案:建立作者智能体(意图转文本)与读者智能体(文本解读与批评)的双智能系统。借鉴叙述交流和接受美学理论,文本意义在作者智能体创作过程中动态生成,通过读者端的“期待视野”填补文本的叙述空白,并影响其叙事判断及反馈。读写智能体双方通过持续反馈循环迭代文本,实现“再叙述”过程,最终达成“意义共振”。该模式实现了创作人的深度意图理解、批判性反馈与高效协同进化,推动人机角色重构:人类转为意图设定与美学指导者,AI成为意图执行伙伴。此范式突破提示词表达困境,为AI创作开辟了深度协作新路径。
关键词:数字文心;AI文学创作;叙述学;大语言模型
作者简介:徐晓雪,深圳大学人文学院硕士研究生,研究方向:数字人文与科幻文学研究。电邮:763341984@qq.com。范力欣(通讯作者),深圳前海微众银行人工智能首席科学家,研究方向:人工智能。电邮:lixin.fan01@gmail.com。
Title: Digital Wenxin: A Narrative Paradigm for Large Models Driven by Meaning Resonance
Abstract: To address the current “mindless” dilemma in AI literary creation, this paper innovatively proposes the Digital Wenxin solution: establishing a dual-agent system comprising an Author Agent (intent-to-text) and a Reader Agent (text interpretation and critique). Drawing upon narrative communication and reception aesthetics theories, textual meaning is dynamically generated during the Author Agent’s creation process. Through the “horizon of expectations” at the reader’s end, narrative gaps in the text are filled, which in turn influences its narrative judgments and feedback. Both the reading and writing agents iterate on the text through a continuous feedback loop to realize a “re-narration” process, ultimately achieving “meaning resonance”. This model facilitates a deep understanding of the creator’s intent, critical feedback, and efficient co-evolution, driving a reconstruction of human-machine roles: humans transition into intent setters and aesthetic guides, while AI serves as an intent execution partner. This paradigm overcomes the expressive limitations of prompt engineering, forging a new path for deep collaboration in AI creation.
Keywords: Digital Wenxin; AI Literary Creation; Narratology; Large Language Models (LLMs)
Author Biographies: Xu Xiaoxue, Master Candidate at the School of Humanities, Shenzhen University. Research Areas: Digital Humanities and Science Fiction Studies. E-mail: 763341984@qq.com. Fan Lixin (Corresponding author), the Principal Scientist of Artificial Intelligence at WeBank, Shenzhen. Research Area: Artificial Intelligence. E-mail: lixin.fan01@gmail.com.
Received: 16 Mar 2026 / Revised: 23 Mar 2026 / Accepted: 11 Apr 2026 / Published online: 30 Apr 2026 / Print published: 30 May 2026.