Multilingual Ontology for Machine Translation is a project carried out by the Cultureplex Lab and Knowtions Research, and supported by SOSCIP. Its mission is to enhance the state of machine translation in terms of procedures and methodologies employed currently. We believe that the best approach is to support machine translation by means of an ontology (that is, a taxonomy of concepts and relationships between them that models the behavior of a particular domain).
The purpose of creating an ontology is to add semantics and make meaning readable by machines, so that the machines can work with contextual relationships in texts that are written in natural language (English, French, Spanish…). Our ontology will be multilingual, domain-specific, concept-based, and language-independent, and we are producing it from different resources such as dictionaries, glossaries, other resources, and parallel corpora (i.e., large collections of original texts and their translations into other languages). The result of this process will be an ontology composed by concepts that link the lexical units of the different languages, which constitutes a powerful tool for translation.
The relevance of this project to Canada is significant. Canada is home to cutting-edge research in translation (the federal government invested approximately $200 million in 2013). We except to augment to state-of-the-art machine translation current engines such as Google’s TensorFlow. This outcome will establish Knowtions Research as the foremost multilingual machine learning and natural language processing company in the Canadian market, where we aspire to occupy a dominant position.