Using neural networks for spelling normalisation

Student: N.N.

Supervisor: Gerold Schneider

Introduction

Texts from earlier stages of English often employ outdated, and non-standardized spelling variants. Rule-based approaches, edit distance, and character-based translation has been used to map the historical variants to present-day English variants. You will review and if possible improve on these approaches, by using state-of-the-art technology, in particular neural networks. As application corpus, we use the ARCHER corpus of English, particularly the sections from 1600-1800.