Spelling Correction as a Foreign Language

TL;DR

This paper presents a novel approach to spelling correction by framing it as a machine translation task using an encoder-decoder neural network, simplifying the process and achieving competitive results without feature engineering.

Contribution

It reformulates spell correction as a machine translation problem with an encoder-decoder neural network, eliminating the need for feature engineering and combining language and error models.

Findings

Competitive performance with state-of-the-art methods

No feature engineering or hand tuning required

Effective use of user log data for training

Abstract

In this paper, we reformulated the spell correction problem as a machine translation task under the encoder-decoder framework. This reformulation enabled us to use a single model for solving the problem that is traditionally formulated as learning a language model and an error model. This model employs multi-layer recurrent neural networks as an encoder and a decoder. We demonstrate the effectiveness of this model using an internal dataset, where the training data is automatically obtained from user logs. The model offers competitive performance as compared to the state of the art methods but does not require any feature engineering nor hand tuning between models.