Exploring the Structure and Sentiment of Twitter Discussions on Online Learning during the Late Pandemic: A Large-Scale Social Network Analysis

Social media platforms provide large-scale, naturally occurring data that enable researchers to examine public discourse and interaction patterns around major societal issues. Focusing on the late pandemic period, this study investigates how Twitter users discussed online learning by combining social network analysis and sentiment analysis. Tweets posted between July 1 and November 28, 2021 were collected using predefined hashtags and search terms in English and Turkish, resulting in a dataset of 6,070,574 tweets. After data cleaning and integration, network structures were modelled through retweet and mention relationships, and key social network metrics (e.g., node and edge counts, average degree, modularity, and network diameter) were computed and visualized using sociograms. In parallel, users’ emotional orientations were examined via dictionary-based sentiment analysis, classifying tweets as positive, negative, neutral, or mixed. Findings show that online learning discussions form clustered interaction patterns that vary by hashtag density and language, while neutral sentiment dominates overall with variations across terms.