arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2012.04332cs.CL
收藏

Facts2Story:通过关键事实控制文本生成

Facts2Story: Controlling Text Generation by Key Facts

Bar Ilan University(巴伊兰大学);Allen Institute for Artificial Intelligence(艾伦人工智能研究所)

Eyal Orbach, Yoav Goldberg

摘要:

自我注意神经网络架构的最新进展提高了开放式文本生成的标准。然而,虽然目前的方法能够产生一个连贯的文本,这是几百个单词长,实现控制的内容,正在生成-以及评估它-仍然是悬而未决的问题。我们提出了一个受控的生成任务,这是基于扩展的事实序列,表示在自然语言,成为一个较长的叙述。我们为这项任务引入了基于人类的评估指标,以及一种用于导出大型训练数据集的方法。我们评估了三种基于微调预训练模型的方法。我们发现,虽然自回归,单向语言模型,如GPT 2产生更好的流畅性,他们努力坚持所要求的事实。我们提出了一个计划和完形填空模型(使用微调XLNet),产生竞争力的流畅性,同时坚持所要求的内容。

英文摘要:

Recent advancements in self-attention neural network architectures have raised the bar for open-ended text generation. Yet, while current methods are capable of producing a coherent text which is several hundred words long, attaining control over the content that is being generated -- as well as evaluating it -- are still open questions. We propose a controlled generation task which is based on expanding a sequence of facts, expressed in natural language, into a longer narrative. We introduce human-based evaluation metrics for this task, as well as a method for deriving a large training dataset. We evaluate three methods on this task, based on fine-tuning pre-trained models. We show that while auto-regressive, unidirectional Language Models such as GPT2 produce better fluency, they struggle to adhere to the requested facts. We propose a plan-and-cloze model (using fine-tuned XLNet) which produces competitive fluency while adhering to the requested content.

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。

↑