Formalize The Semantics Of Natural Language Understanding Research

Natural language understanding is one of the core subjects in artificial intelligence. The essential part of it is to do the research work related to the formalization for various aspects of language problems in computer. Among these, the critical problem is the semantic formalization, which is the most difficult part for the development of the natural language understanding. The main research point for this thesis is to formalize the semantic problems based on linguistic area by using mathematic methods. For the reason that graph, as one mathematic form, is the most similar way to human thinking structure, we build up a platform based on the Knowledge-Graph theory and Conceptual-Graph theory in order to investigate semantic formalization problem.This thesis began its research on natural language understanding skills from the following two aspects: "semantic denotation formalization" and "semantic reasoning formalization". And before it comes to specific formalization methods, it has already done the discussion work on the human cognizing process related to natural language. Because that it is not avoidable to refer to human thinking when it comes to language problems, it is very important, necessary and even decisive to study cognitive science first as the research is mainly about natural language understanding.Firstly, this thesis has mixed the two theories of cognitive, which are "Information-Processing Theory" and "Mental-Model theory", and built a "Cognitive-Model". Then, based on "information" and "mental model", it comes to the discussions about the philosophy, logistic and psychological subjects related to language and semantic field. Based on those, it provides the directions on the concrete methods of language formalization.After that, this thesis comes to the discussion on "semantic denotation formalization problem" based on the Knowledge-Graph theory. Many important conclusions from the previous chapter are applied here. One basic viewpoint is that human beings have a structural expression in their consciousness, which is called mental model. Semantics is the mental model expressed by the language, which is constructed by the concepts. Followed by this, based on Chinese language, the thesis comes to the Knowledge-Graph formalization research related to the four aspects here: concepts, structure, corresponding relations between words and concepts, as well as grammar and structure.