Implementation of Deep Neural Network Algorithm in Developing Chatbot for Automation of Admission and Integrated Service Helpdesk of Universitas Negeri Semarang

Technology can be applied in various aspects of life including in the field of education. One form of technology that can be implemented in the education sector is artificial intelligence, a system created with the aim of enabling machines (computers) to carry out tasks like humans, for example having a conversation. This study aims to apply the Deep Neural Network algorithm and test the accuracy of the algorithm in creating a chatbot as an automation of the Admission and Integrated Service Helpdesk of Universitas Negeri Semarang. The study began with the process of observation and interviews with the Admission and Integrated Service party, data collection, text preprocessing, modeling with the Deep Neural Network algorithm, model evaluation, and deployment in the form of a website. The result, the Deep Neural Network algorithm is worthy of consideration as a classification algorithm in creating a web-based chatbot with model performance reached 100% in all measurement metrics such as accuracy, precision, recall, and F1-score during the training set. On the testing set with new data, the model can classify questions with an accuracy score of 80%, precision 95.24%, recall 83.34%, and F1-score of 88.89%.