Entity Attribute Framework for the Compiling Model of Natural Language Understanding
Honglin Wu, Ruoyi Zhou, Ke Wang · 2017
Semantic knowledge can be separated from language by the knowledge abstract patterns which use entity attribute relationships.With the existing grammar knowledge, the computer can deal with grammar and semantic analysis step by step.Thus the human understanding process is simulated by the computer.This syntax and semantic separation of processing modes is consistent with the compilation of the computer programming language.In this way, we can accurately define the task of understanding natural language by a model of compiling natural language.What we are concerned with is the framework of knowledge, so that computers can use those knowledge accurately.The framework of knowledge include the definition and the organization of the entities, attributes and relationships of the real world.This paper proposed an entity attribute framework for the compiling model of natural language understanding by computer.Language is the carrier of the real world.The entities and attributes in the real world are simple and static objects, which can be directly mapped to the words.We defined the expression system using language units as entity-attribute-framework expression system.In this expression system, we divide the language units into entity words, attribute words, attribute value words and collection frameworks.In order to discuss that the semantic knowledge would have compiled calculation mode, we also compare these knowledge units with the computer programming language in this paper.