Machine Translation of Restaurant Reviews: New Corpus for Domain Adaptation and Robustness

TL;DR

This paper introduces a new French-English restaurant review corpus and a task focused on improving neural machine translation robustness and domain adaptation for user-generated content, with extensive evaluation and novel metrics.

Contribution

It provides a new domain-specific corpus, defines a relevant translation task, and offers baseline models and evaluation metrics tailored for restaurant review translation.

Findings

Significant improvements over existing online translation systems.

Effective baseline models for domain adaptation and robustness.

Proposed task-specific metrics enhance evaluation accuracy.

Abstract

We share a French-English parallel corpus of Foursquare restaurant reviews (https://europe.naverlabs.com/research/natural-language-processing/machine-translation-of-restaurant-reviews), and define a new task to encourage research on Neural Machine Translation robustness and domain adaptation, in a real-world scenario where better-quality MT would be greatly beneficial. We discuss the challenges of such user-generated content, and train good baseline models that build upon the latest techniques for MT robustness. We also perform an extensive evaluation (automatic and human) that shows significant improvements over existing online systems. Finally, we propose task-specific metrics based on sentiment analysis or translation accuracy of domain-specific polysemous words.