Beyond Offline Mapping: Learning Cross Lingual Word Embeddings through Context Anchoring
This work provides a more robust method for learning cross-lingual word embeddings, which is significant for researchers and practitioners working on multilingual natural language processing tasks, especially when traditional mapping methods struggle due to structural differences in monolingual embeddings.
This paper addresses the challenge of aligning cross-lingual word embeddings by proposing a method that learns source language embeddings directly aligned with fixed target language embeddings, rather than mapping two pre-trained spaces. It achieves this by extending skip-gram with translated context words as anchors, outperforming conventional mapping methods on bilingual lexicon induction and showing competitive results on XNLI.
Recent research on cross-lingual word embeddings has been dominated by unsupervised mapping approaches that align monolingual embeddings. Such methods critically rely on those embeddings having a similar structure, but it was recently shown that the separate training in different languages causes departures from this assumption. In this paper, we propose an alternative approach that does not have this limitation, while requiring a weak seed dictionary (e.g., a list of identical words) as the only form of supervision. Rather than aligning two fixed embedding spaces, our method works by fixing the target language embeddings, and learning a new set of embeddings for the source language that are aligned with them. To that end, we use an extension of skip-gram that leverages translated context words as anchor points, and incorporates self-learning and iterative restarts to reduce the dependency on the initial dictionary. Our approach outperforms conventional mapping methods on bilingual lexicon induction, and obtains competitive results in the downstream XNLI task.