Penerapan Metode Adaboost Untuk Multi-Label Classification Pada Dokumen Teks

Abstract

The significant increase in the amount of text data makes the reasons for applying the classification to the text very clear. The manual classification process carried out by humans is very inefficient and ineffective. This limitation opens up great opportunities for the development of automatic text classifications. In the case of article classification, it is more relevant to use multi-label classification, because an article can be categorized into multiple labels. Many approaches can be used to implement a multi-label classification of text. The supervised-learning method in the field of machine learning is a popular method for this problem. In the review conducted, there were journals that carried out a comparative analysis of the supervised method in the multi-label classification. Based on the review conducted, the AdaBoost Algorithm gives better results than other algorithms. Many approaches can be used to implement a multi-label classification of text. The supervised-learning method in the field of machine learning is a popular method for this problem. In the review conducted, some journals carried out a comparative analysis of the supervised method in the multi-label classification. Based on the review conducted, the AdaBoost Algorithm gives better results than other algorithms. Based on research conducted by the AdaBoost algorithm, it gives more optimal results on the dataset with TF-IDF weighting than TF. The results of accuracy, precision, recall, and f-measure given are higher when compared with the comparison algorithm used. The computing time used by the AdaBoost algorithm is faster than the comparison algorithm used.