There are some reasons that our method achieves higher performance against other algorithms. One of these reasons is that our proposed method considered the intrinsic structures of data sets. The samples that are collected onin the real word, usually have intrinsic structures, and our proposed method tries to consideredconsider the intrinsic structures of data when created thecreating hyperplanes.
Also, when we wantwanted to find relevant labels offor a test sample, we examined the membership counting and statistical information of the k nearest neighbor of the test sample. Because, onin real word data, similar labels share the same information and the local information has a greatgreater impact on the performance and efficiency of thea multi-label algorithm. These reasons have caused Thatresulted in the proposed algorithm achieved theachieving higher performancesperformance.