FAWRMT: WITH SPECIAL EMPHASIS ON GRAMMAR DESIGNS AND PARTITIONED PARSING

This paper describes the prototypical MT system, FAWRMT: Fully-automated Weather Reporting Machine Translation. The system is not developed just to replicate the Canadian system of TAUM-METEO. Based on the consensus that FAMT is feasible in a restricted domain such as weather reporting, this project also aims at experimenting with corpus-based statistics and analysis, variated grammar designs and partitioned parsing to enhance the efficiency of the system.