Equidistant k-layer multi-granularity knowledge space

摘要

A multi-granularity knowledge space is a computational model that simulates human thinking and solves complex problems. However, as the amount of data increases, the multi-granularity knowledge space will have a larger number of layers, which will reduce its problem-solving ability. Therefore, we define a knowledge space distance measurement and propose two algorithms to select k representative layers from the multi-granularity knowledge space, where k is specified by the user according to the needs in problem solving, and k representative layers are approximately equidistant. First, we propose a knowledge space distance to measure the distance between any two layers in a multi-granularity knowledge space with superset-subset relationships, and the rationality of the knowledge space distance is verified by theory and experiment. Second, relying on the knowledge space distance and knowledge space distance variance, we propose two algorithms (i.e., a deterministic algorithm and a heuristic algorithm) to select an approximate equidistant k-layer multi-granularity knowledge space. Third, in addition to verifying the effectiveness of the knowledge space distance, the knowledge space distance variance, the deterministic algorithm and the heuristic algorithm, we verify that the equidistant k-layer multi-granularity knowledge space is more efficient than the original multi-granularity knowledge space.