Knowledge Representation in Smart Rules Engine

Paper describes approach to interpretation of the heterogeneous data from sensors based on a new Smart Rules Engine model, that implement cognitive functions as generalization and abstraction of data. Transformation the data from sensors into semantic knowledge as a first stage of data interpretation is introduced. Abstraction and generalization method based on a new model of knowledge representation in the form of multilevel facts. Fuzzy fact characteristic based on fuzzy L-R number is introduced. Smart Rules Engine knowledge representation model takes into account following characteristics of knowledge: the certainty factor, information completeness, aging rate, and dynamics. These characteristics are indicated in the knowledge representation model and are used by the Smart Rules Engine in the knowledge processing by the method of abstraction and generalization. The explainability and adaptability properties of the proposed model is demonstrated in example of interpretation smart machine knowledge about the crossroad situation. It is shown, when semantic knowledge is represented by structured multilevel facts, Smart Rules Engine can meet today's scalability, efficiency, and real time.