Natural Language Generation (NLG) systems in English have been well established for the last two decades. Capabilities have been developed for creating scientific explanations of objects, automated technical documentation, intelligent tutoring systems, story generation, and a number of other application areas. Multilingual generation has been tackled in the last decade, with new systems arriving that produce texts in multiple languages simultaneously from a single knowledge representation. We describe the infrastructure for such a system that produces texts in both English and Italian, where the Italian grammar and the sentence planner were produced in-house at ITC-irst by modifying existing English components. We also briefly describe the overall structure of a multilingual generation system as well as prototype systems we have developed that use these new resources in a variety of ways
Multilingual Generation for Museum Applications
Novello, Alessandra;Callaway, Brendan Charles
2003-01-01
Abstract
Natural Language Generation (NLG) systems in English have been well established for the last two decades. Capabilities have been developed for creating scientific explanations of objects, automated technical documentation, intelligent tutoring systems, story generation, and a number of other application areas. Multilingual generation has been tackled in the last decade, with new systems arriving that produce texts in multiple languages simultaneously from a single knowledge representation. We describe the infrastructure for such a system that produces texts in both English and Italian, where the Italian grammar and the sentence planner were produced in-house at ITC-irst by modifying existing English components. We also briefly describe the overall structure of a multilingual generation system as well as prototype systems we have developed that use these new resources in a variety of waysI documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
