arXivDaily arXiv每日学术速递 周一至周五更新
北美退伍军人节(Veterans Day),12 日无更新
arXiv 1805.03366cs.CL
收藏

基于PU学习的低资源语言的LearningWord嵌入

LearningWord Embeddings for Low-resource Languages by PU Learning

University of Virginia(弗吉尼亚大学);University of California Davis(加利福尼亚大学戴维斯分校);Amazon(亚马逊);University of California Los Angeles(加利福尼亚大学洛杉矶分校)

Chao Jiang, Hsiang-Fu Yu, Cho-Jui Hsieh, Kai-Wei Chang

摘要:

词嵌入是处理自然语言的许多下游应用中的关键组件。现有的方法通常假设存在大量的文本来学习有效的词嵌入。然而,这样的语料库可能不适用于一些低资源语言。在本文中,我们研究如何有效地学习一个词嵌入模型的语料库只有几百万的令牌。在这种情况下,共现矩阵是稀疏的,因为许多词对的共现是未观察到的。与现有的方法通常只采样几个未观察到的词对作为负样本相比,我们认为,共现矩阵中的零条目也提供了有价值的信息。然后,我们设计了一个积极的无标签学习(PU学习)的方法,因式分解的共生矩阵,并在四种不同的语言验证所提出的方法。

英文摘要:

Word embedding is a key component in many downstream applications in processing natural languages. Existing approaches often assume the existence of a large collection of text for learning effective word embedding. However, such a corpus may not be available for some low-resource languages. In this paper, we study how to effectively learn a word embedding model on a corpus with only a few million tokens. In such a situation, the co-occurrence matrix is sparse as the co-occurrences of many word pairs are unobserved. In contrast to existing approaches often only sample a few unobserved word pairs as negative samples, we argue that the zero entries in the co-occurrence matrix also provide valuable information. We then design a Positive-Unlabeled Learning (PU-Learning) approach to factorize the co-occurrence matrix and validate the proposed approaches in four different languages.

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。

↑