Sentiment Analysis on Twitter Using NLP

This study looks at how social media can show what people really think.Still, figuring out emotions in tweets isn't easy -sarcasm and messy text get in the way.We built a tool that uses BERT plus LSTM to detect feelings in posts.To make it clearer why the model decides what it does, we added LIME along with SHAP tools.Our setup works live through Twitter's feed using three parts: one for users, another for crunching data, also one for storing info.It runs on a front end made with React, letting you handle datasets, train models, watch results come in instantly, even see explanations pop up visually.Test results reveal BERT paired with SHAP hits 93% accuracy, topping the rest; meanwhile, LSTM combined with LIME runs quickest while still giving clear explanations.These interpretability tools build user confidence because they make AI choices easier to follow.Instead of guessing why a model decides something, people see the reasons behind it.That openness helps users accept automated outcomes more readily.