Computing Social Presence in Online Discussions Using Natural Language Processing Algorithms: A Conceptual Proposal in Python

In this article, we offer a conceptual proposal for computing social presence scores using Python in online learning environments, specifically in threaded discussion forums, using Term Frequency (TF), Inverse Document Frequency (IDF), and cosine similarity algorithms. We use an atypical approach to applying calculated TF-IDF scores involving frequencies of specific pronouns in individual posts and across the entire corpus of a discussion forum. Finally, we use cosine similarity to compare two vectors: an individual post’s TF-IDF scores per pronoun and the mean TF-IDF scores per pronoun across the corpus. Limitations and recommendations for future research are discussed.