Multi-Label Classification using an Ontology

Abstract

During these last few years, the problem of multilabel classification (ML) has been studied in several domains, such as text categorization. Multi-label classification is a main challenging task because each instance can be assigned to multiple classes simultaneously. This paper studies the problem of Multi-label classification in the context of web pages categorization. The categories are defined in an ontology. Among the weakness of the multi-label classification methods, exist the number of positive and negative examples used to build the training dataset of a specific label. So the challenge comes from the huge number of labels combinations that grows exponentially. In this paper, we present an ontology-based Multilabel classification which exploit dependence between the labels. In addition, our approach uses the ontology to take into account relationships between labels and to give the selection of positive and negative examples in the learning phase. In the prediction phase, if a label is not predicted, the ontology is used to prune the set of descendant labels. The results of experimental evaluation show the effectiveness of our approaches.

Multi-Label Classification of Chinese Rural Poverty Governance Texts Based on XLNet and Bi-LSTM Fused Hierarchical Attention Mechanism

“…In other words, the para process of document-level embeddings are transferred. Be unsuitable for transferring due to the different target catego set Θ ∈ Υ is the parameters set of the classifier in the cl fore, the set of weights | ∈ ( // )× and | ∈ × and parameters of the Bi-LSTM network in Equations (11), can be denoted as ℇ . For other classifier nodes that namely the child nodes of classifier , the parameter set ℇ of the parameter set ℇ for the training of the classifier .…”

“…As the number of labels increases, the parameter size and complexity of the model also significantly increase. In contrast, combining multiple classifiers [11], namely one or a subset of labels corresponding to a classifier, could reduce the complexity of a single model and avoid being influenced by the dataset's quality, yielding a better holistic classification result. Nevertheless, the number of classifiers will increase while the labels are abundant.…”

“…From a process perspective, each text classifier includes four modules: data preprocessing, feature extraction, classification algorithms, and result evaluation [13]. To improve the performance of the classifier chain, researchers typically perform unified preprocessing on all classifier inputs and a unified assessment of the classification results [11]. Feature extraction and classification are two kernel processes that determine the performance of a classifier.…”

“…The metadata-based approaches utilize explicit information such as titles, authors, keywords, categories, and other existing labels in the text as text features. This method is standard for the scientific literature with complete metadata information [11], but for text in poverty governance, metadata information is often not a mandatory configuration. Therefore, content-based feature extraction methods are more valuable in PGT text classification.…”

Transformation of SysML Requirement Diagram into OWL Ontologies

“…Ontology is also able to infer generalization and specialization between classes based on constraints imposed on the property of the class definition [10]. Furthermore, the appropriate concepts of a domain are reflected by the ontology [11]. Therefore, the transformation of the SysML Requirement Diagram into an ontology is needed.…”