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 2302.07912cs.CL
收藏

满足低资源语言的需求:通过预训练模型自动对齐的价值

Meeting the Needs of Low-Resource Languages: The Value of Automatic Alignments via Pretrained Models

University of Colorado Boulder(科罗拉多大学博尔德分校);Johns Hopkins University(约翰斯·霍普金斯大学);University of Edinburgh(爱丁堡大学);Universidad de la República(共和国大学);Northeastern University(东北大学);Universidade Tecnológica Federal do Paraná(巴拉那联邦技术大学);Dartmouth College(达特茅斯学院)

Abteen Ebrahimi, Arya D. McCarthy, Arturo Oncevay, Luis Chiruzzo, John E. Ortega, Gustavo A. Giménez-Lugo, Rolando Coto-Solano, Katharina Kann

摘要:

大型多语言模型激发了一类新的单词对齐方法,这些方法适用于模型的预训练语言。然而,最需要自动对齐的语言是低资源的,因此通常不包括在预训练数据中。在这项工作中,我们要问:现代对齐器在看不见的语言上表现如何,它们比传统方法更好吗?我们为布里布里语-西班牙语、瓜拉尼语-西班牙语、盖丘亚语-西班牙语和Shipibo-Konibo语-西班牙语提供了黄金标准排列。有了这些,我们评估国家的最先进的对齐与不适应目标语言的模型。最后,我们还评估了由此产生的路线外,通过两个下游任务:命名实体识别和词性标注。我们发现,虽然基于变换的方法通常优于传统的模型,这两类方法仍然相互竞争。

英文摘要:

Large multilingual models have inspired a new class of word alignment methods, which work well for the model's pretraining languages. However, the languages most in need of automatic alignment are low-resource and, thus, not typically included in the pretraining data. In this work, we ask: How do modern aligners perform on unseen languages, and are they better than traditional methods? We contribute gold-standard alignments for Bribri--Spanish, Guarani--Spanish, Quechua--Spanish, and Shipibo-Konibo--Spanish. With these, we evaluate state-of-the-art aligners with and without model adaptation to the target language. Finally, we also evaluate the resulting alignments extrinsically through two downstream tasks: named entity recognition and part-of-speech tagging. We find that although transformer-based methods generally outperform traditional models, the two classes of approach remain competitive with each other.

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。

↑