A Maximum Entropy Approach to Natural Language Processing

5.1 Review of statistical translation

1. What was Candide, and how did it translate?
   IBM’s French-to-English statistical MT system: choose \(\hat{E} = \arg\max_E p(E)\, p(F \mid E)\) with a language model and a translation model (word fertilities, word translation probabilities and a distortion model) trained by EM on the Canadian Hansards.
   • Berger et al., 1996, A Maximum Entropy Approach to Natural Language Processing (Computational Linguistics), §5.1
2. What shortcoming of the basic translation model motivated maximum entropy models in Candide?
   It ignores English context when translating a word: it always gives in the same probabilities for dans, pendant… regardless of whether “a month’s time” or “the fiscal year 1992” follows, leading to errors like supérieures → “superior” instead of “greater” before “50%”.
   • Berger et al., 1996, A Maximum Entropy Approach to Natural Language Processing (Computational Linguistics), §5.1