Improving Ensemble Classifier for Natural Language Situational Information Analysis
Mr. Dattatray S. Shingate, Shyamrao V. Gumaste · 2024
Today, Social networking services are being used more popularly, particularly in disaster-like situations. The way we communicate has changed since the advent of social media. It is currently how people receive news. In addition to all these benefits, social media is gradually proving to be a life-saving instrument. Heterogeneous content may be found on many social media platforms, such as Twitter, blogs, news aggregators, etc. During an emergency, a large amount of useful information is published on social media, along with the sympathies and opinions of the people. Therefore, individuals often communicate using their own natural language, which a machine cannot comprehend. In order to share the information that people have about the present situation that is taking place close to them during the crisis, there has to be a reliable approach that is necessary for extracting beneficial information (Context) from the natural language that people generally employ for communication. This information may be further divided into situational and non-situational categories. Situational information highlights current disaster-related effects that call for quick action in order to prevent future losses from occurring. We have proposed a novel method which will be significant over all the existing classifier alone as the prediction will be rely on novel ensemble classifier which will based on multiplicative vectors from LSTM and Robust BERT model.