This document discusses a project to analyze the sentiment levels of timelines posted on Twitter using natural language processing. The project aims to quantify the sentiment levels of posts, create evaluation metrics for casual posts' sentiment levels using NLP, and improve sentiment analysis algorithms and scores. It describes preprocessing tweets obtained from the Twitter API, using morphological analysis, sentiment dictionaries, FastText models trained on Wikipedia, and product review data to score tweets. The results show that current methods do not accurately score slang terms like "uzai", and future work could involve using deep learning to better capture context and extend evaluation.