Fake News Classification Using LSTM Algorithm on Kompas Platform

Fake news or popularly known as “hoax” has become an increasingly disturbing problem in today's digital era. With the rapid development of information technology, the spread of fake news has become easier and faster, threatening the integrity of information presented by leading news platforms such as Kompas. Therefore, this study aims to develop an LSTM (Long Short-Term Memory)-based algorithm in classifying fake news spread through the Kompas platform. The approach used in this study involves an in-depth text analysis process using the LSTM architecture, which is known to be effective in handling sequential data model problems such as text. The dataset used consists of a collection of news that has been labeled whether the news is included in the fake news category or not. Algorithm development is carried out through the stages of text processing, LSTM model creation, and model performance evaluation. This model achieves an accuracy of 98.13% on training data and 94.80% on test data, demonstrating the model's ability to recognize patterns well and generalize to new data.