Pembangunan Aplikasi Chatbot Informasi Akademik berbasis Cosine Similarity dan Library Sastrawi Stemmer (Studi Kasus: Teknik Informatika IT PLN)

Abstract

The fast delivery of accurate academic information for students is one of the problems faced by the Program Studi Teknik Informatika IT PLN. Information received by students from social media belongs to the department, as well as from fellow students is often inaccurate and unclear, or requires further explanation, while questions asked to the department office via chat are not always answered on time. To overcome this problem, a chatbot application was built to produce automatic answers to incoming questions. On the other hand, chat applications, such as WhatsApp, LINE, and Telegram, have provided Application Programming Interface (API) facilities for sending and receiving messages that allow the development of applications to be in the form of chats. Therefore, the Chatbot proposed in this study is implemented on Telegram app and was developed using the Stemmer Literature as a library for the message preprocessing, the TF-IDF method for word weighting, and the Cosine Similarity method for examining similarities between objects. Based on the results of the test, the accuracy is 90% calculated using confusion matrix and the average of cosine similarity