Sentiment Analysis of Restaurant Reviews in Artvin Province by Rule-based Sentiment Analysis and Machine Learning

Abstract

The purpose of this study was to investigate customer sentiments of restaurants in Artvin province. It was determined that 73.9% of the reviews were positive, and 26.1% were negative. 7 topics including place, view, price, food, service, staff and taste were extracted from the reviews. While the most reviews were about the place with 33.89%, it was followed by view with 15%, and the fewest reviews were about taste with 5.83%. It was found that the view topic was the most liked among these topics. 23.53% of those who commented on the price stated that the prices were high, while the percentage of those who indicated that the service was slow was 21.98%. In general, it was noticed that the service, place, food, and view topics were closely related to each other, and a customer who likes one of them is likely to appreciate the others and vice versa. It can be concluded that the application of RBSA and ML methods together is appropriate in terms of enabling both grammar rules and artificial intelligence methods and obtaining satisfactory results.