Learning in computer-mediated setting represents a complex, multidimensional process. This complexity calls for a comprehensive analytical approach that would allow for understanding of various dimensions of learner generated discourse and the structure of the underlying social interactions. Current research, however, primarily focuses on manual or, more recently, supervised methods for discourse analysis. Moreover, discourse and social structures are typically analyzed separately without the use of computational methods that can offer a holistic perspective. This paper proposes an approach that addresses these two challenges, first, by using an unsupervised machine learning approach to extract speech acts as representations of knowledge construction processes and finds transition probabilities between speech acts across different messages, and second, by integrating the use of discovered speech acts to explain the formation of social ties and predicting course outcomes. We extracted six categories of speech acts from messages exchanged in discussion forums of two MOOCs and each category corresponded to knowledge construction processes from well-established theoretical models. We further showed how measures derived from discourse analysis explained the ways how social ties were created that framed emerging social networks. Multiple regression models showed that the combined use of measures derived from discourse analysis and social ties predicted learning outcomes.

Student engagement is often considered an overarching construct in educational research and practice. Though frequently employed in the learning analytics literature, engagement has been subjected to a variety of interpretations and there is little consensus regarding the very definition of the construct. This raises grave concerns with regards to construct validity: namely, do these varied metrics measure the same thing? To address such concerns, this paper proposes, quantifies, and validates a model of engagement which is both grounded in the theoretical literature and described by common metrics drawn from the field of learning analytics. To identify a latent variable structure in our data we used exploratory factor analysis and validated the derived model on a separate sub-sample of our data using confirmatory factor analysis. To analyze the associations between our latent variables and student outcomes, a structural equation model was fitted, and the validity of this model across different course settings was assessed using MIMIC modeling. Across different domains, the broad consistency of our model with the theoretical literature suggest a mechanism that may be used to inform both interventions and course design.

This paper examines the discrete learning strategies employed within a massive open online course and their relationship to the student learning experience. The theoretical framework centered on the Community of Inquiry model of online education, which outlines the three critical dimensions (presences) of student learning experience: teaching, social, and cognitive presence. The Community of Inquiry survey instrument, administered as the part of the post-course survey, was used to measure student perceived levels of the three presences. Cluster analysis revealed three different groups of students with unique study strategies: limited users, selective users, and broad users. The strategies adopted significantly differed in student use of available tools and resources, final course grade, as well as the perceived levels of cognitive presence. The results also indicate there were significant differences regarding student commitment to learning, motivations and goals for enrolling in a MOOC, as well as goal orientation, approaches to learning, and the use of different study strategies. Implications for research and practice of online learning are further discussed.

Automatic question generation, which aims at converting sentences in an article to high-quality questions, is an important task for educational practices. Recent work mainly focuses on designing effective generation architectures based on deep neural networks. However, the first and possibly the foremost step of automatic question generation has largely been ignored, i.e., identifying sentences carrying important information or knowledge that is worth asking questions about. In this work, we (i) propose a total of 9 strategies, which are grounded on heuristic question-asking assumptions, to determine sentences that are question-worthy, and (ii) compare their performance on 4 datasets by using the identified sentences as input for a well-trained question generator. Through extensive experiments, we show that (i) LexRank, a stochastic graph-based method for selecting important sentences from articles, gives robust performance across all datasets, (ii) questions collected in educational settings feature a more diverse set of source sentences than those obtained in non-educational settings, and (iii) more research efforts are needed to further improve the design of educational question generation architectures.

The capacity to foster interpersonal interactions in massive open online courses (MOOCs) has frequently been contested, particularly when learner interactions are limited to MOOC forums. The establishment of social presence-a perceived sense of somebody being present and "real"-is among the strategies to tackle the challenges of online learning and could be applied in MOOCs. Thus far, social presence in MOOCs has been under-researched. Studies that previously examined social presence in MOOCs did not account for the peculiar nature of open online learning. In contrast to the existing work, this study seeks to understand how learners perceive social presence, and the different nuances of social presence in diverse MOOC populations. In particular, we compare perceptions of social presence across the groups of learners with different patterns of forum participation in three edX MOOCs. The findings reveal substantial differences in how learners with varying forum activity perceive social presence. Perceptions of social presence also differed in courses with the varying volume of forum interaction and duration. Finally, learners with sustained forum activity generally reported higher social presence scores that included low affectivity and strong group cohesion perceptions. With this in mind, this study is significant because of the insights into brings to the current body of knowledge around social presence in MOOCs. The study's findings also raise questions about the effectiveness of transferring existing socio-constructivist constructs into the MOOC contexts.

This study presents an evaluation of the Community of Inquiry (CoI) survey instrument developed by Arbaugh et al. (2008) within the context of Massive Open Online Courses (MOOCs). The study reports the results of a reliability analysis and exploratory factor analysis of the CoI survey instrument using the data of 1487 students from five MOOC courses. The findings confirmed the reliability and validity of the CoI survey instrument for the assessment of the key dimensions of the CoI model: teaching presence, social presence, and cognitive presence. Although the CoI survey instrument captured the same latent constructs within the MOOC context as in the Garrison's three-factor model (Garrison et al., 1999), analyses suggested a six-factor model with additional three factors as a better fit to the data. These additional factors were 1) course organization and design (a sub-component of teaching presence), 2) group affectivity (a sub-component of social presence), and 3) resolution phase of inquiry learning (a sub-component of cognitive presence). The emergence of these additional factors revealed that the discrepancies between the dynamics of the traditional online courses and MOOCs affect the student perceptions of the three CoI presences. Based on the results of our analysis, we provide an update to the famous CoI model which captures the distinctive characteristics of the CoI model within the MOOC setting. The results of the study and their implications are further discussed.