Early Classification of Data Streams

Early classification of data streams involves predicting the class label of a data stream before it is fully observed. This approach is particularly valuable in time-sensitive domains, such as the spread of fake news on social media, where information is gathered gradually over time. However, the inherent conflict between accuracy and timeliness poses a significant challenge and requires a solution considering both. In this research, we intend to use learning methods to use an early classification model whose purpose is optimization in this series of problems and measure its performance compared to previous methods. Finally, we will evaluate this model using real data obtained from social networks.