A Semantic Schema - based Approach for Natural Language Translation

摘要

Processing natural language statements to obtain equivalent translations in a different language has long been an area of research in Artificial Intelligence. In machine translation systems, an intermediate representation of inputs is necessary to express the result of the phrase analysis. These representations treat each phrase as a character string and construct the corresponding syntactic and/or semantic representation structure. In the proposed approach, representation is made by means of a semantic network type structure, named semantic schema, with focus on the dependency relations existing between the sentence words. The resulted schema components are further evaluated (using an interpretation system) with the corresponding constructions from the language into which translation occurs.