Abstract
Due to the Coronavirus (Covid-19) pandemic, there was a positive shift in online shopping. On an e-commerce website like Shopee, consumers may post comments under the products they have purchased. This research study aims to conduct sentiment analysis on product reviews as customer recommendations in Shopee Philippines. The product reviews were first scraped from Shopee. After which, it is preprocessed and then annotated using VADER. The customers’ sentiments were analyzed using Multinomial Naive Bayes (MNB) and Support Vector Machine (SVM). The confusion matrix and a classification report were then used to determine the accuracy value, the precision value, the recall value, and the F-measure value of the results from the models. Lastly, the results from a survey would justify the model’s results to customer recommendations. The final results of the research study show how reviews with positive, negative, and neutral sentiments can affect a product’s condition to be recommended to other consumers or not. Based on the analysis of the product reviews, 83.6% are positive, 9.1% are negative, and 7.3% are neutral. The SVM model is found to be a better model than MNB which got an 83% accuracy score. The survey results which validated the model’s results have found that 75.8% of the respondents would recommend a store or a product based on the number of positive reviews.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call