Statistical Machine Translation (SMT) for Highly-Inflectional Scarce-Resource Language

Statistical machine translation (smt) is a machine translation paradigm, in which translations aregenerated on the base of statistical models. in this system, parameters are derived from an analysis of a parallelcorpus, and smt quality depends on the ability of learning word translations. enriching the smt by a suitablemorphology analyser decreases out of vocabulary words and dictionary size dramatically. this could be moreconsiderable when it deals with a highly-inflectional, low-resource, language like persian. defining a suitablegranularity for word segment may improve the alignment quality in the parallel corpus. in this paper differentschemes and word’s combinations segments in a smt’s experiment from persian to english language are prospectedand the best one-to-one alignment, which is called en-like scheme, is proposed. by using the mentioned scheme thetranslation’s quality from persian to english is improved about 3 points with respect to bleu measure over thephrase-based smt.