Um Ambiente para Apoio à Tradução Baseado em Conhecimento : Estudo de Caso com Português-Libras

This dissertation presents a proposal for automatic translation based on knowledge, the design and implementation of a system for authors and use of automated translators to support the translation, and as a case study for the translation of Portuguese to Libras. This system has an environment for manipulation of the elements used in the process of machine translation and one for automatic translation of texts of a language-source, in the form of text, for a target language, in the form of text, video and/or audio. The elements used in the translation process are examples of translation from language-source to the target language and rules of translation inferred considering these examples. Therefore, the content of this paper presents a study on machine translation, pointing their methods and, briefly, its main paradigms, and a brief presentation on memory translation. It also a study on the definition formal language and grammatical inference, pointing their methods and, briefly, the specification of problems of grammatical inference. It also presents the algorithms of translation and inference used by the system as well as the data structures necessary and results generated by them. Throughout the content, it is possible to observe some aspects of usability, navigability, functionality and complexity of the system generated as the final product of this work, an environment to support the translation, based on examples and syntax.