It is undeniable that NMT has been evolving over the last decades. However, the success of this MT technology depends on the availability of large parallel databases for engine training, which entails a significant challenge for many languages. In some cases, there is no bilingual corpora and sometimes it is even hard to find a large monolingual corpus.
Therefore, this webinar seeks to instruct you on the strategies and optimizations that can be done in a low-resource language combination to build and train an NMT engine. Some real examples will be given extrapolating my experience creating NMT resources for the language pair Spanish-Galician.
不可否认,NMT在过去几十年中一直在发展。然而,这种机器翻译技术的成功取决于用于发动机训练的大型并行数据库的可用性,这对许多语言来说是一个重大的挑战。在某些情况下,没有双语语料库,有时甚至很难找到大型的单语语料库。
因此,本次网络研讨会旨在指导您在低资源语言组合中构建和训练NMT引擎时可以执行的策略和优化。我们将给出一些真实的例子来推断我为西班牙语-加利西亚语这对语言创建NMT资源的经验。
以上中文文本为机器翻译,存在不同程度偏差和错误,请理解并参考英文原文阅读。
阅读原文