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 2310.16248cs.CL
收藏

GlotLID:面向低资源语言的语言识别

GlotLID: Language Identification for Low-Resource Languages

LMU Munich(慕尼黑大学);Munich Center for Machine Learning(慕尼黑机器学习中心);Sorbonne Université(索邦大学);CNRS(法国国家科学研究中心);ISIR(信息、信号与系统研究所)

Amir Hossein Kargaran, Ayyoob Imani, François Yvon, Hinrich Schütze

AI总结:

本文提出GlotLID-M语言识别模型,覆盖1665种语言,在F1与假阳性率平衡上优于四个基线,并分析了低资源语言识别的独特挑战。

摘要:

近年来,若干论文已针对约300种高资源和中等资源语言发布了良好的语言识别(LID)解决方案。然而,目前尚无一种LID能够同时满足以下要求:(i)覆盖广泛的低资源语言,(ii)经过严格评估且可靠,(iii)高效且易于使用。本文发布了GlotLID-M,一个满足广泛覆盖、可靠性和效率要求的LID模型。它能够识别1665种语言,与先前工作相比覆盖范围大幅增加。在我们的实验中,在平衡F1和假阳性率(FPR)时,GlotLID-M优于四个基线(CLD3、FT176、OpenLID和NLLB)。我们分析了低资源LID所面临的独特挑战:不正确的语料元数据、来自高资源语言的泄漏、难以区分密切相关的语言、宏语言与变体的处理以及总体上的噪声数据。我们希望将GlotLID-M集成到数据集构建流程中,能够提升低资源语言和文化的NLP技术质量与可及性。GlotLID-M模型(包括未来版本)、代码和数据源列表可在以下地址获取:https://github.com/cisnlp/GlotLID。

英文摘要:

Several recent papers have published good solutions for language identification (LID) for about 300 high-resource and medium-resource languages. However, there is no LID available that (i) covers a wide range of low-resource languages, (ii) is rigorously evaluated and reliable and (iii) efficient and easy to use. Here, we publish GlotLID-M, an LID model that satisfies the desiderata of wide coverage, reliability and efficiency. It identifies 1665 languages, a large increase in coverage compared to prior work. In our experiments, GlotLID-M outperforms four baselines (CLD3, FT176, OpenLID and NLLB) when balancing F1 and false positive rate (FPR). We analyze the unique challenges that low-resource LID poses: incorrect corpus metadata, leakage from high-resource languages, difficulty separating closely related languages, handling of macrolanguage vs varieties and in general noisy data. We hope that integrating GlotLID-M into dataset creation pipelines will improve quality and enhance accessibility of NLP technology for low-resource languages and cultures. GlotLID-M model (including future versions), code, and list of data sources are available: https://github.com/cisnlp/GlotLID.

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。

↑