المستخلص: |
In this paper, we show how Arabic sentences are generated from the Interlingua representations (IRs) used in the KANT knowledge-based machine translation (MT) system, an interlingua-based software architecture for translation from English to several languages [Mitamura 1999, Niberg 1992]. The system which generates Arabic sentences from Interlingua Representations consists of 4 subsystems: the mapping system, the sentence generation system, the sentence/morphology generation interface and the morphological generation system. We describe these subsystems and address some issues in generating Arabic from IRs.
|