Hybrid feature selection model based on machine learning and knowledge graph

Abstract Aiming at the problem that current feature selection algorithm can not adapt to both supervised learning data and unsupervised data, had poor interpretability, this paper proposed a hybrid model based on machine knowledge graph. By idea of hybridization, used algorithms, algorithms graph technology from perspective features text features. Firstly, data-based weights were obtained through model, then text-based by using technology, weight sets are combined obtain matrix with good explanatory properties meets Finally, case analysis proves method in has effects interpretability.