A Conditional Generative Matching Model for Multi-lingual Reply Suggestion
This addresses the challenge of serving multilingual reply suggestion models with skewed data distributions, offering incremental improvements for low-resource languages.
The paper tackles the problem of multilingual automated reply suggestions by proposing a Conditional Generative Matching model (CGM) within a Variational Autoencoder framework, which improves relevance by over 10% on average in ROUGE scores and diversity by 80% compared to baselines.
We study the problem of multilingual automated reply suggestions (RS) model serving many languages simultaneously. Multilingual models are often challenged by model capacity and severe data distribution skew across languages. While prior works largely focus on monolingual models, we propose Conditional Generative Matching models (CGM), optimized within a Variational Autoencoder framework to address challenges arising from multi-lingual RS. CGM does so with expressive message conditional priors, mixture densities to enhance multi-lingual data representation, latent alignment for language discrimination, and effective variational optimization techniques for training multi-lingual RS. The enhancements result in performance that exceed competitive baselines in relevance (ROUGE score) by more than 10\% on average, and 16\% for low resource languages. CGM also shows remarkable improvements in diversity (80\%) illustrating its expressiveness in representation of multi-lingual data.