Metagraph Representation for Overcoming Limitations of Existing Knowledge Bases

Currently, knowledge bases are actively used in many applications, such as search engines, dialog assistants. Primarily, it happens due to the improvement of existing methods through the analysis of meaning. Knowledge bases have different knowledge representations. Semantic networks, frames, semantic web, production rules are used for this purpose. But existing methods of knowledge representation have limitations. Thus, RDF is a flat model, and production rules contradict each other. Therefore, it is proposed to apply metagraphs to combine the advantages of different representations. The paper describes the emergence feature of this method for complex knowledge representation. An analytical assessment of the search time is carried out for a concept in the knowledge base. The authors implemented the proposed method of representation in a knowledge base management system and carried out an experiment to evaluate its performance.