SENTIMENT ANALYSIS OF CUSTOMER SATISFACTION TOWARDS STEAK HUT MANYAR KERTOARJO RESTAURANT SERVICES USING THE TF-IDF METHOD

On online shopping sites or often referred to as marketplaces, there is a column of comments and reviews of transactions that have been made by buyers for products that have been purchased. With this product assessment feature, buyers can consider decisions about the products they will buy. But at this time there is a problem with the review feature because many buyers give negative comments but give a five-star rating. This results in the feature of giving values from consumers being bad. For this reason, a sentiment analysis study was conducted on the review feature at the Steakhut Manyar restaurant using the naive Bayes method and the Tf-Idf algorithm. Based on the review of reviews at the Steakhut restaurant, 1000 review data have been collected which are divided into two, namely 700 training data and 300 test data. After that, the text preprocessing data stage is carried out, where the text preprocessing stage is collecting product and service review data on the web page (Cleaning data), changing uppercase letters to lowercase letters (Casefolding), separating sentences into single sentences (tokenizing), removing conjunctions that are not used for sentiment analysis (stopwords), changing words to basic words (stemming) and continuing to give weight to each word using the Tf-idf algorithm