arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2206.11219cs.CL
收藏

理解生成语料库的性质

Understanding the Properties of Generated Corpora

IBM Research(IBM研究院)

Naama Zwerdling, Segev Shlomov, Esther Goldbraich, George Kour, Boaz Carmeli, Naama Tepper, Inbal Ronen, Vitaly Zabershinsky, Ateret Anaby-Tavor

摘要:

文本生成模型已成为许多研究任务的焦点,特别是对于句子语料库的生成。然而,理解自动生成的文本语料库的属性仍然具有挑战性。我们提出了一套工具,检查生成的文本语料库的属性。将这些工具应用于各种生成语料库,使我们能够对生成模型的属性有新的认识。作为表征过程的一部分,我们发现两种领先的生成技术生成的语料库存在显着差异。

英文摘要:

Models for text generation have become focal for many research tasks and especially for the generation of sentence corpora. However, understanding the properties of an automatically generated text corpus remains challenging. We propose a set of tools that examine the properties of generated text corpora. Applying these tools on various generated corpora allowed us to gain new insights into the properties of the generative models. As part of our characterization process, we found remarkable differences in the corpora generated by two leading generative technologies.

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。

↑