The purpose of this research is to perform sentiment analysis on Twitter data using Natural Language Processing (NLP) techniques, particularly leveraging the NLTK library in Python within a Jupyter notebook environment. The study aims to explore sentiment classification methods, evaluating the emotional tone of tweets and categorizing them as neutral, positive, or negative sentiments, utilizing NLTK's SentimentIntensityAnalyzer. The sample consists of Twitter data with columns like 'Tweet' and 'Sentiment' sourced from a CSV file. The methodology involves tokenizing and processing the text, grading sentiment, counting occurrences of the hashtag #fifa, and analyzing word frequencies [1]. In addition to the lexicon-based VADER approach, the study incorporates a transformer-based deep learning model—BERT (Bidirectional Encoder Representations from Transformers) -to enhance sentiment classification accuracy. BERT, pre-trained on large corpora and capable of understanding context and nuanced language, offers a state-of-the-art alternative to traditional models. This inclusion allows a comparative analysis between rule-based and deep learning approaches, highlighting BERT’s effectiveness in handling complex tweet structures. Furthermore, the study investigates the impact of removing stopwords and explores the list of eliminated stopwords. The expected results include gaining insights into prevalent sentiments on Twitter regarding a specified topic, frequency of the hashtag #fifa, and a comprehensive understanding of word usage, visually depicted through wordclouds. Possible limitations include inherent subjectivity in sentiment analysis, potential variations in language use, reliance on hashtag frequency as an indicator of topic prevalence, and the effectiveness of stopword removal, which may be context-dependent. The addition of wordcloud analysis enhances the visual representation of the most frequent words, providing a holistic perspective on the dataset.

Sentiment analysis Twitter data mining VADER BERT NLP