arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2008.07466cs.CL
收藏

故事生成与理解的叙事插值

Narrative Interpolation for Generating and Understanding Stories

UT Austin(德克萨斯大学奥斯汀分校)

Su Wang, Greg Durrett, Katrin Erk

摘要:

我们提出了一种控制叙事/故事生成的方法,在这种方法中,我们能够引导模型通过插值生成具有用户指定目标结局的连贯叙事:例如,我们被告知Jim去远足,最后Jim需要被救出,我们希望模型在此过程中逐步生成步骤。我们的方法的核心是基于GPT-2的插值模型,该模型以叙事中的前一句和下一句为条件,并填补空白。此外,reranker有助于控制生成的文本的连贯性。与人类的评价,我们表明,结束指导的生成结果的叙述是连贯的,忠实于给定的结束指南,并需要更少的人工努力的一部分,人类指南作家比过去的方法。

英文摘要:

We propose a method for controlled narrative/story generation where we are able to guide the model to produce coherent narratives with user-specified target endings by interpolation: for example, we are told that Jim went hiking and at the end Jim needed to be rescued, and we want the model to incrementally generate steps along the way. The core of our method is an interpolation model based on GPT-2 which conditions on a previous sentence and a next sentence in a narrative and fills in the gap. Additionally, a reranker helps control for coherence of the generated text. With human evaluation, we show that ending-guided generation results in narratives which are coherent, faithful to the given ending guide, and require less manual effort on the part of the human guide writer than past approaches.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑