Hospitals as health service institutions need to adapt to community needs to provide
comfortable and quality services. Hospital accreditation, including sentiment
analysis of public opinion, is important in assessing patient satisfaction. Therefore,
this research aims to analyze public sentiment towards JIH Purwokerto Hospital
services using the Naïve Bayes Classifier method. Data obtained through web
scraping of Google Maps reviews. The data used in this research were 247 reviews
using three categories, namely "doctor services", "pharmacy services" and "cashier
services". The collected data then goes through a series of preprocessing stages,
including cleaning the data from less meaningful elements or words. After
preprocessing, the data is labeled using a Lexicon Based approach. The feature
extraction stage is carried out using TF-IDF, Bigram and Trigram word weighting
techniques to increase classification accuracy. The data is divided into training
data and test data with a ratio of 80:20, where the Naïve Bayes Classifier algorithm
is used to classify sentiment based on the extracted features. The evaluation results
show that the Naïve Bayes algorithm with TF-IDF feature extraction is the best with
an accuracy of 87%, a precision value of 81%, a recall value of 75% and an f1-
score value of 77%.