Dynamic Updating the Three-Way Regions for the Optimal Rules

Knowledge acquisition is an important research hotspot in the field artifical intelligence. Most of the research methods mainly mine all decision rules from the static data. However, they face two aspects: (1) Some old possible rules with lower confidences may be useless; (2) Some optimal rules from the new data are unable to be effectively obtained. How to acquire the optimal rules from dynamic data while ensuring efficiency becomes an important challenge. To this end, a dynamic updating mechanism for the optimal rules using three-way decisions is designed. More specifically, the incremental updating strategies for three-way regions are proposed and illustrated when multiple objects are dynamically changed from the decision table. Secondly, the two corresponding incremental learning algorithms are implemented for the dynamic objects using three-way decisions. Finally, a series of experiments on eight data sets are validated our models. Experiment results demonstrate that our proposed incremental updating algorithms for the optimal rules can significantly accelerate the knowledge updating.