Multilingual Distributed Representations without Word Alignment

TL;DR

This paper introduces a method for learning multilingual distributed word representations without requiring word alignments, enabling cross-lingual semantic tasks and outperforming previous methods on classification benchmarks.

Contribution

It presents a novel approach to learn shared multilingual embeddings directly from sentence alignments without word-level alignments, capturing cross-lingual semantics.

Findings

Outperforms previous state-of-the-art in cross-lingual document classification

Learns semantic relationships across languages without parallel data

Produces semantically informative multilingual representations

Abstract

Distributed representations of meaning are a natural way to encode covariance relationships between words and phrases in NLP. By overcoming data sparsity problems, as well as providing information about semantic relatedness which is not available in discrete representations, distributed representations have proven useful in many NLP tasks. Recent work has shown how compositional semantic representations can successfully be applied to a number of monolingual applications such as sentiment analysis. At the same time, there has been some initial success in work on learning shared word-level representations across languages. We combine these two approaches by proposing a method for learning distributed representations in a multilingual setup. Our model learns to assign similar embeddings to aligned sentences and dissimilar ones to sentence which are not aligned while not requiring word alignments. We show that our representations are semantically informative and apply them to a cross-lingual document classification task where we outperform the previous state of the art. Further, by employing parallel corpora of multiple language pairs we find that our model learns representations that capture semantic relationships across languages for which no parallel data was used.