Unsupervised Cross-lingual Word Embedding by Multilingual Neural Language Models
This addresses the challenge of cross-lingual NLP for researchers and practitioners by providing an unsupervised method that works with limited or mismatched data, though it is incremental as it builds on existing neural language model techniques.
The authors tackled the problem of obtaining cross-lingual word embeddings without parallel data by proposing multilingual neural language models with shared LSTMs, achieving higher quality embeddings than existing unsupervised models, especially with small or domain-mismatched monolingual data (e.g., 50k sentences).
We propose an unsupervised method to obtain cross-lingual embeddings without any parallel data or pre-trained word embeddings. The proposed model, which we call multilingual neural language models, takes sentences of multiple languages as an input. The proposed model contains bidirectional LSTMs that perform as forward and backward language models, and these networks are shared among all the languages. The other parameters, i.e. word embeddings and linear transformation between hidden states and outputs, are specific to each language. The shared LSTMs can capture the common sentence structure among all languages. Accordingly, word embeddings of each language are mapped into a common latent space, making it possible to measure the similarity of words across multiple languages. We evaluate the quality of the cross-lingual word embeddings on a word alignment task. Our experiments demonstrate that our model can obtain cross-lingual embeddings of much higher quality than existing unsupervised models when only a small amount of monolingual data (i.e. 50k sentences) are available, or the domains of monolingual data are different across languages.