Currently, obtaining information has become easier with the advent of internet technology. Online news portals provide on-demand access to desired information. However, the abundance of news content can make it difficult to find specific articles due to manual categorization errors. This research focuses on evaluating the performance of the Deep Learning method using a Recurrent Neural Network (RNN) for multi-classification tasks on news headlines related to Economics, Health, Sports, and Politics. Training and testing data were collected from news portals using Web Scraping, followed by Text Preprocessing stages such as case folding, tokenization, stopwords removal, and stemming. TF-IDF was then used for feature extraction to assign weights to each word. Testing the model's performance using the Confusion Matrix showed an accuracy of 97%, indicating that the RNN method effectively handles news headline classification and can be applied in news classification systems.
Copyrights © 2023