An Innovative Method of Developing BI-LSTM and KNN Algorithms for Disintegrating Sentiments Through Reviews

During the COVID-19 pandemic, schools globally transitioned to online education due to lockdowns, prompting concerns about its acceptance and effectiveness among stakeholders. This study analyzes public sentiment towards e-learning using a dataset of 17,155 tweets collected from social media platforms like Facebook, Instagram, and Twitter. Machine learning and deep learning techniques are employed for text analysis, including sentiment analysis using Senti Word Net, Text Blob, and VADER. Various machine learning models, supported by feature extraction methods like Bag of Words (BoW) and TF-IDF, are used to categorize sentiment with measures such as accuracy, precision, recall, and F1 score. Long short term memory (LSTM), k nearest neighbour (KNN) machine classifiers, when combined with BoW data, achieve the highest accuracy. Subject modeling reveals common e-learning challenges, including insufficient network infrastructure, children’s difficulties with online learning, and uncertainties about campus reopening.