Abstract
This paper describes the building and evaluation of a translation table using the von\r Mises-Fisher directed statistical distribution. We propose three models to create a directional\r source to target language translation table. We evaluate two of these translation tables against\r a baseline multinomial statistical translation table built using the Giza++ open source\r software package. The best source to target word performance using our model is 6% in the\r best match category, which is the same as baseline accuracy. The benefits of a directed\r statistical approach over a traditional translation table are leveraged translation models on\r minimal data and an abundance of close-match word translations.