Beyond Contrastive Learning: A Variational Generative Model for Multilingual Retrieval
This work addresses the challenge of efficient and effective multilingual retrieval for applications like cross-lingual question answering, though it appears incremental as it builds on existing generative and contrastive methods.
The paper tackles the problem of learning multilingual text embeddings for retrieval tasks by proposing a generative model that separates semantic information from language-specific variation, and it shows that this model outperforms strong contrastive and generative baselines on tasks like semantic similarity and bitext mining.
Contrastive learning has been successfully used for retrieval of semantically aligned sentences, but it often requires large batch sizes or careful engineering to work well. In this paper, we instead propose a generative model for learning multilingual text embeddings which can be used to retrieve or score sentence pairs. Our model operates on parallel data in $N$ languages and, through an approximation we introduce, efficiently encourages source separation in this multilingual setting, separating semantic information that is shared between translations from stylistic or language-specific variation. We show careful large-scale comparisons between contrastive and generation-based approaches for learning multilingual text embeddings, a comparison that has not been done to the best of our knowledge despite the popularity of these approaches. We evaluate this method on a suite of tasks including semantic similarity, bitext mining, and cross-lingual question retrieval -- the last of which we introduce in this paper. Overall, our Variational Multilingual Source-Separation Transformer (VMSST) model outperforms both a strong contrastive and generative baseline on these tasks.