Formalizations of error models with applications to spelling error correction

Abstract

For many information processing applications, there are several different existing error models and error correction algorithms. This research focuses on a general methodology for defining error models describing different types of errors in information processing. It includes formal definitions of channels and the error models and a general algorithm for applying an error model to correct errors. This general methodology represents all existing error models and corrects errors in a consistent way.

This research also discusses the computation of error models with application to spelling error correction. Different error models for various spelling error correction problems have been investigated. The improved Brill and Moore error model has been implemented to describe the approach of computing a spelling error model for specific users. Based on the general methodology devised in this research, four error models based on the improved Brill and Moore error model have also been described and tested.