Machine translation based on unification link grammar

摘要

The goal of this article is to develop a translation method based on link grammar. Through this and to make it possible, we introduce a new formalism for syntactic analysis called Unification Link Grammar as a result of adding unification paradigms to link grammar. This brings about some other benefits such as more generative power, less complexity and being well organized for knowledge extraction. Based on this new formalism, we develop a translation method which is completely compatible with the linguistic basis of link grammar. This approach transfers graphs by using some kind of correspondence between the nodes and without using so called sub-constituents, sub-graphs or any similar concept. The introduced algorithm has a linear order time complexity. The implemented system shows the advantages of our approaches in covering complex structures of English language.