Arquitectura de desambiguación lingüística guiada por dominios de conocimiento distribuidos para traducción automática de lengua española a LESCO

Abstract

Current statistical methods necessity of large sets of phrase examples between two languages and the increasing research in sign languages, implies the search of new strategies to make more flexible inclusion of new vocabulary and produce it correctly according to the speech context. This thesis introduces the design and implementation of a distributed architecture of semantic threads for a Spanish language to Costa Rican Sign Language (LESCO) translation. It is showed the effectiveness of a contextual recognition algorithm proposed for this architecture through tests of informative texts and the use of linguistically validated corpora.