Utilizing Character and Word Embeddings for Text Normalization with Sequence-to-Sequence Models

TL;DR

This paper presents a sequence-to-sequence model utilizing character and word embeddings, including subword information, to improve text normalization in languages with limited annotated data, achieving state-of-the-art results in Arabic.

Contribution

It introduces a novel approach combining character-based attention with pre-trained word embeddings that incorporate subword information for effective text normalization.

Findings

Achieved state-of-the-art F1 score on Arabic text normalization dataset.

Demonstrated effectiveness of subword-aware embeddings in low-resource language tasks.

Showed that combining character and word embeddings enhances neural models for NLP.

Abstract

Text normalization is an important enabling technology for several NLP tasks. Recently, neural-network-based approaches have outperformed well-established models in this task. However, in languages other than English, there has been little exploration in this direction. Both the scarcity of annotated data and the complexity of the language increase the difficulty of the problem. To address these challenges, we use a sequence-to-sequence model with character-based attention, which in addition to its self-learned character embeddings, uses word embeddings pre-trained with an approach that also models subword information. This provides the neural model with access to more linguistic information especially suitable for text normalization, without large parallel corpora. We show that providing the model with word-level features bridges the gap for the neural network approach to achieve a state-of-the-art F1 score on a standard Arabic language correction shared task dataset.