Universidad de Alicante (SPAIN)
About this paper:
Appears in: INTED2019 Proceedings
Publication year: 2019
Pages: 8718-8722
ISBN: 978-84-09-08619-1
ISSN: 2340-1079
doi: 10.21125/inted.2019.2175
Conference name: 13th International Technology, Education and Development Conference
Dates: 11-13 March, 2019
Location: Valencia, Spain
Traditionally, the communication between students and teachers has been based on face-to-face and virtual tutorials, as well as, FAQS. The virtual tutorials allow online asynchronous conversations between students and teachers, which, in general, includes a time delay. The FAQs allow to answer student questions quickly by showing top and frequent questions.

Recently, conversational agents in the form of chatbots have become a reality on social media and messaging apps thanks to the advances on artificial intelligence. Although conversational agents exist since 1965, its term was coined for the first time in 1994 by Michael Mauldin (creator of the first Verbot Julia). Today, chatbots are designed to execute tasks as simple as making the customer purchase easier, or as complex as giving health, technical or legal advice. The use of chatbots is widespread in online businesses as virtual assistants, increasing the confidence of the clients by ensuring better customer service 24 hours a day, seven days a week.

Regarding the application of TICs in an educational context, and the particular case of UACloud at the University of Alicante, the application of chatbots in teaching would facilitate the work of teachers and provide immediate answers to frequently asked questions and virtual tutorials of the students.

The chatbot prototype presented in this paper has being trained through a set of tutorials facilitated by several professors from different subjects in the Computer Engineering and Multimedia Engineering Degrees. Through the analysis of the tutorials, the different conversational flows instances have been designed; the training phrases have been extracted and classified according to the type of tutorial; finally, the different intents have been established.

We observed that tutorials are a rich source of information to train an educational chatbot. The analysis of the text of the tutorials shows that in a high percentage the same questions are repeated by different students. Finally, we discuss the potential of chatbots in an educational context, identifying gaps and future work.
Chatbot, Artificial Intelligence, Natural Language Processing, Tutorials.