Urdu - Roman Transliteration via Finite State Transducers
Tina Bögel · KOPS (University of Konstanz) · 2012
This paper introduces a two-way Urdu–Roman transliterator based solely on a nonprobabilistic finite state transducer that solves the encountered scriptural issues via a particular architectural design in combination with a set of restrictions. In order to deal with the enormous amount of overgenerations caused by inherent properties of the Urdu script, the transliterator depends on a set of phonological and orthographic restrictions and a word list; additionally, a default component is implemented to allow for unknown entities to be transliterated, thus ensuring a large degree of flexibility in addition to robustness.