Transliteration and Longest Mathch Strategy

TRANSLITERATION AND LONGEST MATCH STRATEGY

Dimiter Skordev

Abstract: A natural requirement on transliteration systems is that transliteration and its inversion could be easily performed. To make this requirement more precise, we consider a text transduction as easily performable if it can be accomplished by a finite transducing device such that all successful tokenizations of input words are compliant with the left-to-right longest-match strategy. Applied to inversion of transliteration this gives a convenient sufficient condition for reversibility of transliteration.

Keywords: left to right, longest match, transliteration, reversible transliteration, sequential transducer.

ACM Classification Keywords: E.4 Data: Coding and information theory – Formal models of communication