Research On Semantic Understanding Model Based On Image Schema

In the field of Natural language processing, semantic representation have been the difficulty and focus of the study.The research has important theoretical significance for many applications in the field of natural language processing, such as information retrieval, machine learning, text generation, question answering and etc.There are difficulties.in semantic formalization, include:the depth of reasoning can’t be predicted, the bottom line of formal atomic propositions is difficult.to be defined, conceptualization causes semantic granularity tends to be fixed and etc. The occurrences and solutions of many language problems are involving the intrinsic properties of entire speech mechanism. The research of semantic representation should go beyond the classic linguistics, computational linguistics areas, and make cognitive science, especially cognitive linguistics as base.Image schema is the main object of study of cognitive linguistics.The image schema itself does not require any explanation. It itself has the meaning, its structure having a logical. When it is activated, it can correspond to the category or concepts. It is the most important semantic structure. The schematheory provides a range of sememes closer to the human underlying level cognitive. Using the image schemas to represent the relations between objects can support semantic-based reasoning more effectively.The main work of this paper is to propose a method of semantic representation based on the image schema. On this basis, build a model of natural language semantics understanding to achieve the representation of the concept of the base layer and an understanding of simple statements.The main work and innovative results of the dissertation is organized as follows.Firstly, introducetheresearchstatusoftheoverseasandthedomestic.Alsoexplaintheobjectandthegoaloft hework. The dissertationis trying to find the way to represent semantic closer to the real representation, and on this basis to build a language semantic understanding model.Secondly, based on the analyzing of image schema theory and prototype theory, the dissertation proposes a new semantic system model for simulating human cognitive system, s and on this basis divided semantic into three elements as relationship between objects, the object’s attributes and object, given the analysis of the relationship between the three elements and the modules within semantics system model.Among them, the image schema used to represent relationships between objects.The representation of the visual image schemas which are simplifies and play a fundamental role in image schema theory is researched, and semantic understanding algorithm based on the visual image schema is proposed.On this basis, a more general semantic representation method based on image schemas is explored. Give a semantic understanding algorithm combined with relationship between objects, the object’s attributes and object.Finally, according to the theory has been researched, thedissertation designed and implemented a semantic understanding model system based on image schemas. An example proves that the proposed semantic representation and language semantic understanding model is feasible.