DOC: Improving Long Story Coherence With Detailed Outline Control
Kevin Yang, Dan Klein, Nanyun Peng, Yuandong Tian
摘要
We propose the Detailed Outline Control (DOC) framework for improving long-range plot coherence when automatically generating several-thousand-word-long stories. DOC consists of two complementary components: a detailed outliner and a detailed controller. The detailed outliner creates a more detailed, hierarchically structured outline, shifting creative burden from the main drafting procedure to the planning stage. The detailed controller ensures the more detailed outline is still respected during generation by controlling story passages to align with outline details. In human evaluations of automatically generated stories, DOC substantially outperforms a strong Re 3 baseline (Yang et al., 2022) on plot coherence (22.5% absolute gain), outline relevance (28.2%), and interestingness (20.7%). Humans also judged DOC to be much more controllable in an interactive generation setting. 1. Daisy is a kind-hearted old woman. She has cancer. Bill is her husband. 2. Lisa is Daisy's daughter. Structured Prompt For Drafting Daisy is diagnosed with cancer. Lisa is trying to find a viable treatment. Lisa has been stressed out lately, and Daisy expresses her concern. Lisa tirelessly continues her research. Lisa finally finds a cure. Setting: Lisa's laboratory. Lisa looked back at Daisy, her eyes clear and full of determination. "I've got this, mom. Hang in there.
问问这篇 Paper
智能体会读完全文。
Lune 把这篇 Paper 索引到了每一个公式,引用它的顶会 Paper 也一样。你提问,回答直接引用原文。
引用它的顶会 Paper27
- Generating Images with Multimodal Language ModelsJing Yu Koh, Daniel Fried, Russ SalakhutdinovNeurIPS 2023 · 被引用 403 次
- Art or Artifice? Large Language Models and the False Promise of CreativityTuhin Chakrabarty, Philippe Laban, Divyansh Agarwal, Smaranda Muresan 等CHI 2024 · 被引用 122 次
- Synthetic Prompting: Generating Chain-of-Thought Demonstrations for Large Language ModelsZhihong Shao, Yeyun Gong, Yelong Shen, Minlie Huang 等ICML 2023 · 被引用 98 次
- LayoutPrompter: Awaken the Design Ability of Large Language ModelsJiawei Lin, Jiaqi Guo, Shizhao Sun, Zijiang Yang 等NeurIPS 2023 · 被引用 71 次
- Rehearsal: Simulating Conflict to Teach Conflict ResolutionOmar Shaikh, Valentino Emil Chai, Michele Gelfand, Diyi Yang 等CHI 2024 · 被引用 62 次
相关 Paper
- Re3: Generating Longer Stories With Recursive Reprompting and RevisionKevin Yang, Yuandong Tian, Nanyun Peng, Dan KleinEMNLP 2022 · 被引用 77 次
- Collective Critics for Creative Story GenerationMinwook Bae, Hyounghun KimEMNLP 2024 · 被引用 3 次
- PlotMachines: Outline-Conditioned Generation with Dynamic Plot State TrackingHannah Rashkin, Asli Celikyilmaz, Yejin Choi, Jianfeng GaoEMNLP 2020 · 被引用 100 次
- Content Planning for Neural Story Generation with Aristotelian RescoringSeraphina Goldfarb-Tarrant, Tuhin Chakrabarty, Ralph M. Weischedel, Nanyun PengEMNLP 2020 · 被引用 106 次
- IS-CoT: Breaking the Long-form Generation Collapse via Interleaved Structural ThinkingZechen Sun, Yuyang Sun, Zecheng Tang, Juntao Li 等ACL 2026
