A web - based application for language translation using semantic web

Abstract

The paper is a report of our work in the thesis project in two semesters: one for research and the other for implementation. Our scopes of this thesis are we understand the Semantic Web technology and apply it to a real, helpful system. We desire to build a web-based Vietnamese-English machine translator (VEMT). The system is constructed by Example-based Machine Translation (EBMT) method and use theWeb Ontology Language (OWL) for consulting. The reasons for this combination are Semantic Web implies a necessary standard annotation that is hopefully going to be used by the whole WWW community. RDF/OWL notation provides additional meaning, and machine translation, EBMT in particular, can make use of these additional annotations for several purposes. As RDF/OWL model aims at enriching documents with information about their content, it can help the Example-Based machine translation. Also there, architecture for the extraction of translation correspondences was proposed. VEMT is a web-based application which simply has an input as a Vietnamese sentence and the result is an English sentence. The system begin with looking for the input sentences in a database, this part is in process of EBMT. If there no match, then the input is divided into words; each word is assigned to its part-of-speed (POS) tagger. Those words are reordered according to rules of translation and POS taggers. After reordering, the ontology engine does its work as a dictionary, matches Vietnamese words to English words. Finally, the system arranges those words to sentence and output this English sentence. Regardless the above ideas, Semantic Web and Machine Translation themselves are not new, but the combination of those two content is new, especially in Vietnam. We believe that our project is a new step to solve one of the hardest problems from the beginning of computer age: machine translation.