A Simple and Effective Approach to Robust Unsupervised Bilingual Dictionary Induction
This work addresses the problem of zero accuracy in unsupervised bilingual dictionary induction for distant language pairs, which is a significant incremental improvement for researchers and practitioners working with low-resource language pairs.
Existing unsupervised bilingual dictionary induction methods fail for distant language pairs due to a gap between actual and minimum required initialization performance. This work proposes Iterative Dimension Reduction to bridge this gap, achieving 13.64-55.53% accuracy for English with Chinese, Japanese, Vietnamese, and Thai, without impacting similar language pairs.
Unsupervised Bilingual Dictionary Induction methods based on the initialization and the self-learning have achieved great success in similar language pairs, e.g., English-Spanish. But they still fail and have an accuracy of 0% in many distant language pairs, e.g., English-Japanese. In this work, we show that this failure results from the gap between the actual initialization performance and the minimum initialization performance for the self-learning to succeed. We propose Iterative Dimension Reduction to bridge this gap. Our experiments show that this simple method does not hamper the performance of similar language pairs and achieves an accuracy of 13.64~55.53% between English and four distant languages, i.e., Chinese, Japanese, Vietnamese and Thai.