HYBRID SENTIMENT ANALYSIS OF MAXIM APP USERS USING SUPPORT VECTOR MACHINE AND LEXICON-BASED APPROACH

Abstract

Online transportation applications such as Maxim have rapidly grown alongside technological advancements. These platforms accumulate large volumes of user reviews on sources like the Google Play Store, providing valuable insights into user perceptions. However, the unstructured nature of textual data makes systematic analysis difficult. This study proposes a sentiment classification model to categorize Maxim user reviews into positive and negative sentiments, excluding neutral responses. The method integrates a lexicon-based approach using the InSet Lexicon with a Support Vector Machine (SVM) classifier. Preprocessing steps included text cleaning, case folding, normalization, tokenization, stopword removal, and stemming. Feature extraction was conducted using Term Frequency–Inverse Document Frequency (TF-IDF), followed by sentiment classification with SVM. Evaluation using a confusion matrix achieved an accuracy of 96.07%. For negative sentiment, the model obtained a precision of 79%, recall of 83%, and F1-score of 81%; for positive sentiment, precision was 89%, recall 98%, and F1-score 93%. These results indicate that integrating lexical resources with machine learning provides an effective solution for sentiment analysis of user-generated reviews.