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 1901.02646cs.CL
收藏

语言表征真正代表什么?

What do Language Representations Really Represent?

University of Copenhagen(哥本哈根大学);Stockholm University(斯德哥尔摩大学);University of Zurich(苏黎世大学);University of Helsinki(赫尔辛基大学)

Johannes Bjerva, Robert Östling, Maria Han Veiga, Jörg Tiedemann, Isabelle Augenstein

摘要:

在文本语料库上训练的神经语言模型可以用于诱导单词的分布式表示,使得相似的单词最终具有相似的表示。如果语料库是多语言的,则可以使用相同的模型来学习语言的分布式表示,使得类似的语言最终具有类似的表示。我们表明,即使多语言语料库已被翻译成英语,通过拿起微弱的信号留下的源语言,这也成立。然而,就像在单词表示中分离语义和句法相似性是一个棘手的问题一样,语言表示捕获的相似性类型并不明显。我们调查的相关性和因果关系的语言表征从翻译学习一方面,遗传,地理,和其他语言之间的结构相似性的几个层次。其中,结构相似性被发现与语言表征相似性相关性最强,而遗传关系-在以前的工作中用于评估的一个方便的基准-似乎是一个混杂因素。除了对翻译效果的影响外,我们更普遍地认为这是一个NLP和语言类型学可以相互作用并相互受益的案例。

英文摘要:

A neural language model trained on a text corpus can be used to induce distributed representations of words, such that similar words end up with similar representations. If the corpus is multilingual, the same model can be used to learn distributed representations of languages, such that similar languages end up with similar representations. We show that this holds even when the multilingual corpus has been translated into English, by picking up the faint signal left by the source languages. However, just like it is a thorny problem to separate semantic from syntactic similarity in word representations, it is not obvious what type of similarity is captured by language representations. We investigate correlations and causal relationships between language representations learned from translations on one hand, and genetic, geographical, and several levels of structural similarity between languages on the other. Of these, structural similarity is found to correlate most strongly with language representation similarity, while genetic relationships---a convenient benchmark used for evaluation in previous work---appears to be a confounding factor. Apart from implications about translation effects, we see this more generally as a case where NLP and linguistic typology can interact and benefit one another.

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。

↑