amLite: Amharic Transliteration Using Key Map Dictionary

TL;DR

amLite is a framework that converts ASCII transliterations of Amharic back into original Amharic script, enhancing data consistency and interoperability for researchers, with high accuracy demonstrated on test data.

Contribution

the paper introduces a novel transliteration framework using a key map dictionary to accurately revert ASCII Amharic transliterations to original script.

Findings

97.7% accuracy on test data

99.7% accuracy on test data

98.4% accuracy on test data

Abstract

amLite is a framework developed to map ASCII transliterated Amharic texts back to the original Amharic letter texts. The aim of such a framework is to make existing Amharic linguistic data consistent and interoperable among researchers. For achieving the objective, a key map dictionary is constructed using the possible ASCII combinations actively in use for transliterating Amharic letters; and a mapping of the combinations to the corresponding Amharic letters is done. The mapping is then used to replace the Amharic linguistic text back to form the original Amharic letters text. The framework indicated 97.7, 99.7 and 98.4 percentage accuracy on converting the three sample random test data. It is; however, possible to improve the accuracy of the framework by adding an exception to the implementation of the algorithm, or by preprocessing the input text prior to conversion. This paper outlined the rationales behind the need for developing the framework and the processes undertaken in the development.