Research On Construction Of Entity-Attributes-Framework Semantic Knowledge Base

The research purpose of natural language processing is how to make use of computing to understand and generate natural language, so the research on this field has been exploring and trying various language models and processing mechanism to do computing with language. According to the processing approach, the current natural language processing research can be divided into two schools of rule-based rationalist and statistical based empiricist. However, with the rapid development of information-based society, information processing, especially language text processing has made higher demand on the research and application of natural language processing. Nevertheless, all present processing methods are stayed in analyze superficial layer of language which has been becoming a bottleneck for raising the processing precision and achieving language understanding. Therefore, more and more researchers began to consider introducing essential semantic analysis into language understanding process. For this purpose, the research on construction of semantic knowledge base for natural language understanding has become a hot topic.However, in the course of study on construction of semantic knowledge base, researchers just blindly reference semantics research results to design the structure of semantic knowledge base. Since they had not a precision definition of language understanding in computing mode and didn’t put forward a rational abstraction of the structure of semantic knowledge, the current knowledge base that mixed with grammar knowledge has a poor coupling with computing, and semantic knowledge units are scattered. In addition, the existing construction approach of knowledge base relies on experts manual editing. This method is divorced from real language using context, and has defects on completeness.To sum up, this paper deeply analyzes the process of human how to understand their language, and make an abstract of the core structure of semantic knowledge which played an important role in language understanding. We put forward a semantic knowledge system which consisted with entities-attributes-framework lexical knowledge and situation descriptive framework. In order to give the priority of constructing the entity-attributes-framework, we followed the way that human learned world by classification and gave a plan of using real corpus to extract knowledge resources by fields. We used computer field as example to introduce how to use current technology in natural language processing to abstract the knowledge resources automatically. We extracted a certain scale of knowledge about computer entity, attributes and relations, and use these resources to construct an example knowledge base. The methods that we used in knowledge extraction have a good performance in experiment, and can be extended to other fields in their entity-attributes-framework knowledge constructing.