Classification based on label semantic characteristic analysis
Mingwei Zhang, Ziqi Ji, Zebo Dong · 2017
Classification is a machine learning technique that assigns labels to a collection of data in order to aid in more accurate predictions and analysis. It has been widely studied and many classification approaches have been proposed from different fields. However, there is no one optimal classifier in all respects or for all the data sets as No-Free-Lunch theorem states. In addition, there is a lack of attention to do classification by analyzing label semantic characteristics which can reflect the intrinsic characteristics of one label distinguishing from the others. So from the perspective of knowledge modeling and mining about label semantic characteristics, new classification idea was put forward and a concrete label characteristic value oriented classification approach was proposed in this paper. Experimental results demonstrated that the proposed classification algorithm has relatively high training performance and high accuracy, and classification based on label semantic characteristic analysis is an effective classification idea.