A Novel Deconversion Algorithm for English to Tamil Translation using Universal Networking Language (UNL)

This paper proposes a novel algorithm for sentence formation in Tamil Language from the Enconverted UNL Knowledge graph. Universal Networking language is an intermediary representation of the natural language in the form of graphs. UNL was invented to avoid barriers in Language Translation. The system is fed with children's stories in English, which were enconverted into UNL graphs. In this paper, we focus on the deconverter module and propose a novel algorithm for rearranging the sentences according to the available Tamil sentence forming rule base. The rules are framed in such a way that the sentences follow Subject- Object- Verb (SOV) format. Also, the rules ensure that the original meaning is conveyed correctly after translating from English to Tamil. The Translation quality of the proposed system is evaluated with various performance metrics such as BLEU, TER and METEOR. The use of UNL in translation gives more flexibility and adaptability in handling various linguistic structures and nuances. Overall, this approach shows promise for improving cross-lingual communication and text processing tasks.