Information service centers are very important today, with the existence of information service centers many people are helped in conveying and receiving the information needed, including in the world of higher education, information about academics is needed in teaching and learning activities in lectures. Although to get general information related to current teaching and learning activities at the Faculty of Information Technology at Budi Luhur University can be accessed through the website, supervisors and social media. But the problem now is to get information from supervisors who sometimes take too long to respond to students or even miss not responding. One way to overcome this problem is to create a chatbot system as an information service at the Faculty of Information Technology at Budi Luhur University using a natural language Natural Language Processing (NLP) approach with the Neural Network method and Multilayer Perceptron algorithm and extraction features using the binary Bag of words method by matching and giving a value of 1 to each question that is used as a token at the appropriate preprocessing stage on The train data and assigns a value of 0 to each toen that does not match the token on the train data. As well as using datasets saved in JSON format. Based on the model that has been trained to obtain accuracy and loss results with an accuracy value = 1,000 and a loss value = 0.0117, it can be concluded that the trained model is a good model.