Artificial Intelligence in Education: An automatic Rule-Based Chatbot to generate guidance from lecture recordings

Hing, William; Gordon, Neil; Al Jaber, Tareq

Abstract

In a new era of educational and research-based chatbots, implementing personalised interactive learning resources is critical in enhancing students' academic experiences. Whilst general purpose chatbots are now available with a range of platforms, there are concerns about the nature of the content, and on the resources required to run the. This research involves an innovative integration of Artificial Intelligence (AI) within an educational context, to develop an advanced bespoke chatbot based on transcriptions and summaries of lecture recordings. This novel tool represents an evolution from passive to active learning resources, likely to improve student learning engagement and comprehension [2]. Leveraging advanced Natural Language Processing (NLP) techniques, this chatbot aims to foster an engaging learning environment by transforming passive lecture content into an interactive, query-answering interface, whilst enabling some measure of how much resource such platforms require. The chatbot employs several key AI components: Speech recognition for transcriptions, Name Entity Recognition (NER) for topic and keyword extraction, Spacy for processing user queries, Wordnet for synonym generation and language identification and translation to handle non-English queries. Furthermore, the chatbot uses TF-IDF for information retrieval and the T5 transformer model for summarisation which is renowned for its semantic comprehension and ability to produce context-aware responses. This allows the chatbot to provide detailed lecture summaries and define complex terms using API integration. User engagement is facilitated using PyQt5 to develop a user-friendly graphical interface. The interface offers a variety of features, including adjustable text size, a theme switch feature to transition between light and dark modes, multiple conversation management and conversation deletion options. These features enhance the user's overall experience and allow personalised chatbot interaction. While there are challenges related to the accuracy of transcriptions, topic modelling reliability and the quality of responses, this project aims to impact the future of personalised, interactive, and educational resources.

Citation

Hing, W., Gordon, N., & Al Jaber, T. (2024). Artificial Intelligence in Education: An automatic Rule-Based Chatbot to generate guidance from lecture recordings. Acta Scientific Computer Sciences, 6(7), 64-74

• Journal Article Type: Article
• Acceptance Date: Jun 21, 2024
• Online Publication Date: Jul 1, 2024
• Publication Date: Jul 1, 2024
• Deposit Date: Jun 21, 2024
• Publicly Available Date: Jul 23, 2024
• Journal: Acta Scientific Computer Sciences
• Publisher: Acta Scientific
• Peer Reviewed: Peer Reviewed
• Volume: 6
• Issue: 7
• Pages: 64-74
• Keywords: Artificial Intelligence; ChattBot; Technology Enhanced Learning
• Public URL: https://hull-repository.worktribe.com/output/4715824
• Publisher URL: https://actascientific.com/ASCS/ASCS-06-0539.php