Research And Application Of Hierarchical Multi-label Classification Algorithm

In classification tasks,an object usually belongs to one class within a set of disjoint classes.In more complex classification tasks,an object can belong to more than one class,in what is conventionally termed multi-label classification.Moreover,there are cases in which the set of classes are organised in a hierarchical fashion,and an object must be associated to a single path in this hierarchy,defining the so-called hierarchical classification.Finally,in even more complex scenarios,the classes are organised in a hierarchical structure and the object can be associated to multiple paths of this hierarchy,defining the problem investigated in this article: hierarchical multi-label classification.This paper presents two classification algorithms to solve this problem.One is hierarchical multi-label classification based on path selection,which trains a multi-class classifier for each parent node.In the classification phase,the branches with low probability to occur are pruned,performing non-mandatory leaf node prediction.This method evaluates each possible path from the root of the hierarchy,taking into account the prediction value and the level of the nodes: selecting the path(or paths)with the score higher than a given threshold.The other is hierarchical multi-label classification based on neural network.The algorithm trains a neural network model for each layer of the hierarchical label tree separately,and links these neural network models into a neural network chain as the final prediction model.Meanwhile,the algorithm splices the output of each layer of neural network as the final output.Finally,we investigate the selection of both single and multiple thresholds to predict the final label set.In this paper,the hierarchical multi-label classification algorithm is applied to large-scale network intelligent operation and maintenance ticket classification tasks,and a hierarchical multilabel classification system based on B\S architecture has been developed.The actual operation results show that this system can realize hierarchical multi-label classification,and effectively visualize multi-label classification tree and multi-label classification results.