RECONTRA-categorized

Abstract

Encouragingly accurate translations have recently been obtained using a connectionist translator called RECONTRA (Recurrent Connectionist Translator). In contrast to traditional Knowledge-Based systems, this model is built from training data resulting in an Example-Based approach. It directly carries out the translation between the source and target language and employs a simple (recurrent) connectionist topology and a simple training scheme. This paper extends previous work exploring the capabilities of this RECONTRA model to perform text-to-text translations in limited-domain tasks.