Classifying News Based on Indonesian News Using LightGBM
Arif Ridho Lubis, Santi Prayudani, Yulia Fatmi, Okvi Nugroho
Abstract
There were several news categories that present editors with challenges. Some news categories, such megapolitan, national, celebrity, news, lifestyle, and economics, used vocabulary that was quite similar to that of the other categories. International required an editor to be familiar with the article's contents in order for it to be uploaded and placed in the proper category. We had to first digest the news before we could label it compared to other kinds of data. The text mining approach, which attempts to make text or documents may be processed so that it will aid in the process of news classification, will be used to categorize and determine the type of news in this context. The Light Gradient Boosted Machine (LightGBM) model was used in this study to increase the gradient point with a learning stage and obtain the optimal value. This model's training process was intended to be quick while consuming less storage space and processing information more accurately. The accuracy of the classifications made using a confusion matrix to quantify the findings of this investigation, which were news type classifications, was 86%.