Live, Die, Evaluate, Repeat: Do-Over Simulation in the Generation of Coherent Episodic Stories

Live, Die, Evaluate, Repeat: Do-Over Simulation in the Generation of Coherent Episodic Stories
复制标题

生、死、评估、重复:在生成连贯的情节故事中进行重复模拟

DOI:
--
复制
发表时间:
2018
期刊:
International Conference on Innovative Computing and Cloud Computing
影响因子:
--
通讯作者:
T. Veale
T. Veale
中科院分区:
--
文献类型:
--
作者:
Stefan Riegl;T. Veale

文献摘要

被引文献

相似文献

苦苦挣扎的作家有时会忍不住放弃他们目前的努力,从一张白纸和一种新的方法重新开始。但屈从于这种诱惑的代价是高的,特别是当一个人已经在一项必须丢弃的工作上投入了大量的时间和精力的时候。然而,随着计算创造力提高了叙事生成的速度并降低了成本,重新开始的选择变得越来越有吸引力。因此,我们在这里考虑一种基于模拟的方法来生成插曲故事,其中故事在快速、粗粒度的参与和评价(fl)循环中被生成、评估和频繁丢弃。模拟的目的是为了更好地开发故事中人物之间信息传递的情景可能性,而重复模拟的目标是寻找(fi)和实现情节部分之间最大连贯的故事。
Struggling writers are sometimes tempted to throw away their current effort and start over with a a blank page and a fresh approach. But the cost of yielding to this temptation is high, especially when one has already sunk a great deal of time and energy into a work that must be discarded. However, as computational creativity increases the speed and lowers the cost of narrative generation, the option of a fresh do-over becomes ever more attractive. So we consider here a simulation-based approach to the generation of episodic stories in which stories are generated, evaluated and frequently discarded in a rapid, coarse-grained cycle of engagement and reflection. The goal of simulation is to better exploit the situated possibilities for information transfer amongst the characters in a story, while the goal of repeated simulation is to find the story that achieves maximal coherence amongst its episodic parts.