Semantic Network as a Knowledge Base in Training Systems
Bessmertnyy, Igor A.
Kulagin, Viacheslav S.
MetadataShow full item record
Training system (TS) is a subclass of expert systems, i.e. knowledge based information systems. Knowledge representations in TS may differ depending on their purpose and the topic. Simple TS implement computerized questionnaires and the knowledge base may consist of questions and answers. TS based on expert system shells use facts and rules, like all the expert systems do. The paper presented describes an approach to TS building where knowledge presented by semantic network. Semantic network as a knowledge base has a lot of advantages: syntax and language independence, visibility, simplicity of inserting new information. The item of semantic network is a triple including a subject, an object and the predicate that defines the relation between them. Additional dictionaries help to eliminate problem of synonymy and polysemy and let arrange a simple natural language interface because the program has to match just a user three-word sentence with a triple in the semantic network. The Program "Semantic", the first release of TS based on these principles, written in Visual Prolog, supports a verbal human-computer dialog and displays the growing graph of facts established during the conversation. The program is included to educational process in St.Petersburg State University of Informational Technologies, Mechanics and Optics.