<![CDATA[In this study, a chatbot for new student admissions was developed using the Naïve Bayes algorithm. The data source used was a dialogue between admissions staff and prospective students, which was based on information from the new student admissions brochure of Institut Widya Pratama (formerly STMIK Widya Pratama) for the 2024-2025 academic year. The data was arranged in JSON format, including three elements: question structure, classification, and answers. Data pre-processing steps included text normalization, label encoding, vectorization, and word embedding. The Naïve Bayes algorithm was chosen to train the model because of its ability to predict responses from discrete data. The test results showed that this algorithm achieved an accuracy level of 88%, surpassing the performance of the K-Neighbors-Classifier algorithm. The resulting model was implemented in a desktop-based chatbot application that was run via the command prompt, using the Python programming language and several libraries such as pickle, string, and numpy. This application is able to provide answers to user questions by accessing the model stored in the *.pkl file. The results of the study show that the Naïve Bayes algorithm is effective in predicting user questions, with sufficient accuracy for a new student admissions chatbot application.]]>
Copyrights © 2025
