%0 Generic %1 jiang2018learningword %A Jiang, Chao %A Yu, Hsiang-Fu %A Hsieh, Cho-Jui %A Chang, Kai-Wei %D 2018 %K naacl2018 session6 wordembeddings %T LearningWord Embeddings for Low-resource Languages by PU Learning %U http://arxiv.org/abs/1805.03366 %X 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.

@misc{jiang2018learningword, abstract = {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.}, added-at = {2018-06-03T19:21:44.000+0200}, author = {Jiang, Chao and Yu, Hsiang-Fu and Hsieh, Cho-Jui and Chang, Kai-Wei}, biburl = {https://www.bibsonomy.org/bibtex/2894e5e70588f65c726a006c870612f73/albinzehe}, description = {[1805.03366] LearningWord Embeddings for Low-resource Languages by PU Learning}, interhash = {0b71b9415972453b66eabd13b9bd7fdc}, intrahash = {894e5e70588f65c726a006c870612f73}, keywords = {naacl2018 session6 wordembeddings}, note = {cite arxiv:1805.03366Comment: Published in NAACL 2018}, timestamp = {2018-06-03T19:21:44.000+0200}, title = {LearningWord Embeddings for Low-resource Languages by PU Learning}, url = {http://arxiv.org/abs/1805.03366}, year = 2018 }