Chatbot for Complex Questions in University Admission using Bidirectional Long-Short Term Memory and Convolutional Neural Network

Setiabudi, Djoni Haryadi and Budhi, Gregorius Satia and SAMPURNO, ALFONS RICHARDO (2024) Chatbot for Complex Questions in University Admission using Bidirectional Long-Short Term Memory and Convolutional Neural Network. [UNSPECIFIED]

	PDF Download (367Kb)
	PDF Download (1543Kb)
Preview	PDF (paper - gregorius) Download (3318Kb) | Preview

Abstract

Petra Christian University already has services to answer questions about new student admissions through WhatsApp, email, and Instagram. However, given the many inquiries, not all can be answered quickly, especially the complex and lengthy questions. There has been much research on chatbots for new student admissions, but the chatbots developed have not yet examined how well they can answer complex and lengthy questions. Therefore, this research will contribute by investigating how accurately the chatbot can answer short, medium, and long questions using Bidirectional Long-Short Term Memory and Convolutional Neural Network methods. The results show that the chatbot can answer short questions with the highest accuracy of 93%, but its accuracy drops below 80% when given medium and complex questions.

Actions (login required)