arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2004.14373cs.CLcs.LG
收藏

ToTTo:一个受控的表格到文本生成数据集

ToTTo: A Controlled Table-To-Text Generation Dataset

Google Research(谷歌研究院);Carnegie Mellon University(卡内基梅隆大学);Georgia Tech(佐治亚理工学院)

Ankur P. Parikh, Xuezhi Wang, Sebastian Gehrmann, Manaal Faruqui, Bhuwan Dhingra, Diyi Yang, Dipanjan Das

摘要:

我们提出了ToTTo,一个开放领域的英语表到文本数据集,有超过120,000个训练样本,提出了一个控制生成任务:给定一个维基百科表和一组突出显示的表单元格,产生一个句子的描述。为了获得生成的目标是自然的,但也忠实于源表,我们引入了一个数据集的建设过程中,注释者直接修改现有的候选句子从维基百科。我们对我们的数据集和注释过程进行了系统的分析,以及通过几个最先进的基线实现的结果。虽然通常很流畅,但现有的方法经常会出现表格不支持的短语,这表明该数据集可以作为高精度条件文本生成的有用研究基准。

英文摘要:

We present ToTTo, an open-domain English table-to-text dataset with over 120,000 training examples that proposes a controlled generation task: given a Wikipedia table and a set of highlighted table cells, produce a one-sentence description. To obtain generated targets that are natural but also faithful to the source table, we introduce a dataset construction process where annotators directly revise existing candidate sentences from Wikipedia. We present systematic analyses of our dataset and annotation process as well as results achieved by several state-of-the-art baselines. While usually fluent, existing methods often hallucinate phrases that are not supported by the table, suggesting that this dataset can serve as a useful research benchmark for high-precision conditional text generation.

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。

↑