Morphological and Language-Agnostic Word Segmentation for NMT

TL;DR

This paper compares linguistically uninformed and motivated subword segmentation methods for neural machine translation, finding that non-motivated methods currently perform better, but a simple preprocessing step can improve BPE results.

Contribution

It introduces a novel derivational dictionary-based segmentation method and demonstrates how preprocessing can enhance BPE performance in NMT.

Findings

Non-motivated methods outperform linguistically motivated ones in German-Czech translation.

A simple preprocessing step significantly improves BPE translation quality.

Identifies a key difference between BPE and STE methods.

Abstract

The state of the art of handling rich morphology in neural machine translation (NMT) is to break word forms into subword units, so that the overall vocabulary size of these units fits the practical limits given by the NMT model and GPU memory capacity. In this paper, we compare two common but linguistically uninformed methods of subword construction (BPE and STE, the method implemented in Tensor2Tensor toolkit) and two linguistically-motivated methods: Morfessor and one novel method, based on a derivational dictionary. Our experiments with German-to-Czech translation, both morphologically rich, document that so far, the non-motivated methods perform better. Furthermore, we iden- tify a critical difference between BPE and STE and show a simple pre- processing step for BPE that considerably increases translation quality as evaluated by automatic measures.