Several years ago, I provided an introduction to the role of learning analytics in the context of the Community of Inquiry framework and how we might facilitate collaborative learning through the automatic monitoring of discourse and tracking the progress of learners (Editorial 14). I concluded that “learning analytics has enormous potential to help educators understand the characteristics of critical discourse as well as identify areas after the fact that could be improved through redesign”. Furthermore, I noted that these “preliminary results indicate the potential of merging learning analytics with the CoI framework”. The effective application of learning analytics requires a sound theoretical framework. Since then another study (Hu, Mello, & Gašević, 2021) has offered another method for automatic analysis of cognitive presence online discussion messages “using deep machine learning and explainable artificial intelligence” (p. 1). To be clear, while the focus wason analytical methods, the ultimate goal is to provide “more information to support decision-making from instructors in relation to the students’ participation in the online discussion (p. 11). The bottom line, however, is that to make learning analytics an effective tool to support collaborative learning, automatic classification must be extended to the other presences.

In this regard, my primary goal here is to review a recent article that explores a broad range of features for the automatic detection of social presence in an online community of inquiry (Andre, Ferreira Mello, Nascimento, Dueire Lins, & Gašević, 2022). Their approach outperformed previous automatic analyses of social presence classifying the affective, interactive, and cohesive categories (Zou, 2021). They note the practical implications of their study is relevant to studying CoI presences, to assess learner interaction, and facilitate the tracking of learners over time. This study moves us closer to the automatic detection of the dynamics of social presence and understanding the effective use of learning analytics to monitor and manage a community of inquiry to maximize the potential of collaborative inquiry.

As before, this post is simply intended to bring attention to the important work being done regarding learning analytics and convey the potential for automatic classification of discourse in a community of inquiry. Hopefully this will encourage others to develop the tools that will ultimately make is more efficient and effective for practitioners to monitor and manage each of the presences in a community of inquiry. The missing piece, however, is teaching presence. The obvious next step is to initiate work on the analysis of teaching presence as well as detecting shared metacognition (self and shared regulation).