Two Approaches to Correcting Homophone Confusions in a Hybrid Machine Translation System
Pierrette Bouillon, Johanna Gerlach, Ulrich Germann, Barry Haddow, Manny Rayner · Archive ouverte UNIGE (University of Geneva) · 2013
In the context of a hybrid French-to-English SMT system for translating on-line forum posts, we present two meth-ods for addressing the common problem of homophone confusions in colloquial written language. The first is based on hand-coded rules; the second on weighted graphs derived from a large-scale pro-nunciation resource, with weights trained from a small bicorpus of domain language. With automatic evaluation, the weighted graph method yields an improvement of about +0.63 BLEU points, while the rule-based method scores about the same as the baseline. On contrastive manual evalua-tion, both methods give highly significant improvements (p < 0.0001) and score about equally when compared against each other. 1 Introduction and