Triangulation of Reordering Tables: An Advancement Over Phrase Table Triangulation in Pivot-Based SMT

Triangulation in Pivot-Based Statistical Machine Translation(SMT) is a very effective method for building Machine Translation(MT) systems in case of scarcity of the parallel corpus. Phrase Table Triangulation helps in such a resource constrained setting by inducing new phrase pairs with the help of a pivot. However, it does not explore the possibility of extracting reordering information through the use of pivot. This paper presents a novel method for triangulation of reordering tables in Pivot Based SMT. We show that the use of a pivot can help in extracting better reordering information and can assist in improving the quality of the translation. With a detailed example, we show that triangulation of reordering tables also improves the lexical choices a system makes during translation. We observe a BLEU score improvement of 1.06 for Marathi to English MT system with Hindi as a pivot, and also significant improvements in 8 other translation systems by using this method.