Lexicon-Based Sentiment Analysis for Movie Review Tweets

Sentiment analysis is a computational process to identify and classify subjective information such as positive, negative and neutral from the source material. It is able to extract feeling and emotion from a piece of a sentence. This technology has been widely used to extract valuable information from people's views on social media. Hence, this project aims to classify movie reviews into positives, negatives and neutral polarity using lexicon-based method which used R as the language and development framework. Twitter data is used as the source material. Firstly, tweets were extracted using RStudio and Twitter API. Then data pre-processing was done by removing all the stop words and noises. Next was the tokenization process, which separates the words and matches the separated words with positive and negative words vocabulary. Finally, the result of the sentiment analysis is produced into positive, negative and neutral polarities. The results were evaluated using standard evaluation metrics that are the precision, recall, F1 score and accuracy. After all, it is found that the basic lexicon-based method is able to classify sentiment quite well with 52% accuracy. Apparently, the accuracy value achieved in our experiment is not impressive enough, but it is worth corresponding to the simplicity and minimal cost of development for sentiment analysis on Twitter data for movies.