Work

Design of an AI-Driven Feedback and Decision Analysis in Online Learning with Google BERT

Year: 2024
Type: article
Abstract: The global COVID-19 pandemic has significantly altered educational practices. The enforcement of social distancing rules led to the widespread closure of schools, prompting a shift towards remote and online learning modalities. This transition has been challenging for both educators and students. Teachers have struggled to create and deliver online content that meets student needs, while students have faced difficulties adapting to new technologies and resource constraints. The pandemic has also disrupted traditional academic schedules, delaying admission processes, examinations, and academic calendar events. Research is underway to understand the impact of these shifts on student performance and educational outcomes. Interestingly, the pandemic has highlighted the necessity for greater investment in teacher training programs and digital infrastructure to support distance learning. This study introduces an automated feedback assessment model that utilizes Google's Bidirectional Encoder Representations from Transformers (BERT). The model generates a quality score for inputs in a Virtual Learning Environment (VLE) during the pandemic. It was trained using a dataset comprising 10,000 feedback entries, categorized as either "good" or "bad". Further refinement was done on the Open University Learning Analytics (OULA) dataset across 50 epochs. The model achieved a 93.4% accuracy rate on the validation set, indicating its proficiency in evaluating the quality of feedback. The implications of this model are far-reaching. It can be applied in various sectors, including education, performance assessment, and customer service, offering a means to decrease the time and subjectivity involved in human evaluations. This study not only addresses the immediate challenges posed by the pandemic in the educational sector but also provides a forward-looking solution with versatile applications.
Language: English
Open Access status: diamond
OpenAlex ID: W7216814900
FWCI: 5.153
Cites: 26
Cited by: 30
Related to: 10