Rule-Based Sentiment Analysis for Twitter(X) Data

Sentiment analysis on social media data, particularly Twitter, plays a crucial role in understanding public opinion and user sentiments towards various topics. This paper presents a rule-based approach to sentiment analysis tailored for Twitter data, denoted as Twitter(X). The system incorporates a set of positive and negative words relevant to the context of Twitter conversations. A Python implementation using the Natural Language Toolkit (NLTK) showcases the application of this rule-based sentiment analysis on a sample dataset of Twitter(X) posts. The sentiment analysis algorithm considers the frequency of positive and negative words in each tweet, providing a classification of sentiment as positive, negative, or neutral. While the rule-based approach offers simplicity and interpretability, it acknowledges the challenges associated with linguistic nuances, negation, and context-dependent sentiments in natural language. The proposed method utilizes predefined rules and a basic tokenization strategy to assess the sentiment of tweets. In this paper, four rule-based sentiment analyses are used those are Flair, Textblob, Vader, and the transfer, learning model. Transfer learning gives the highest accuracy in terms of sentiment scores for the Twitter data. The paper concludes with discussions on the limitations of the rule-based approach and suggests potential avenues for enhancing sentiment analysis accuracy on diverse and dynamic social media datasets.