Deep Natural Language Processing for Italian Sign Language Translation

This paper presents the architecture of a translator from written Italian into Italian Sign Language. We describe the main features of the four modules of this architecture, i.e. a dependency parser for Italian, an ontology based semantic interpreter, a generator based on expert-systems and combinatory categorial grammars, a planner to position signs in space. The result of this translation chain is signed by a virtual character. Finally, we report the results of a first "intrinsic" experiment for the evaluation of translation quality.