A Conditional Generative Matching Model for Multi-lingual Reply Suggestion

TL;DR

This paper introduces Conditional Generative Matching models within a Variational Autoencoder framework to improve multilingual reply suggestion systems, addressing data skew and capacity challenges, and demonstrating significant performance gains.

Contribution

The paper proposes a novel Conditional Generative Matching model for multilingual reply suggestions, incorporating expressive priors and latent alignment to enhance performance across languages.

Findings

Outperforms baselines with over 10% higher ROUGE scores on average.

Achieves 16% improvement in low-resource languages.

Enhances diversity by 80%, demonstrating better data representation.

Abstract

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.