Despite the benefits of MOOCs (e.g., open access to education offered by prestigious universities), the low level of student engagement remains as an important issue causing massive dropouts in such courses. The use of reward-based gamification strategies is one approach to promote student engagement and prevent dropout. However, there is a lack of solid empirical studies analyzing the effects of rewards in MOOC environments. This paper reports a between-subjects design study conducted in a MOOC to analyze the effects of badges and redeemable rewards on student retention and engagement. Results show that the implemented reward strategies had not significant effect on student retention and behavioral engagement measured through the number of pageviews, task submissions, and student activity time. However, it was found that learners able to earn badges and redeemable rewards participated more in gamified tasks than those learners in the control group. Additionally, results reveal that the participants in the redeemable reward condition requested and earned earlier the rewards than those participants in the badge condition. The potential implications of these findings in the instructional design of future gamified MOOCs are also discussed.

Accurate link quality predictions are key in community wireless mesh networks (CWMNs) to improve the performance of routing protocols. Unlike other techniques, online machine learning algorithms can be used to build link quality predictors that are adaptive without requiring a predeployment effort. However, the use of these algorithms to make link quality predictions in a CWMN has not been previously explored. This paper analyses the performance of 4 well-known online machine learning algorithms for link quality prediction in a CWMN in terms of accuracy and computational load. Based on this study, a new hybrid online algorithm for link quality prediction is proposed. The evaluation of the proposed algorithm using data from a real large scale CWMN shows that it can achieve a high accuracy while generating a low computational load.

Abstract. This paper discusses

No abstract

Predicting the decrease of students' engagement in typical MOOC tasks such as watching lecture videos or submitting assignments is key to trigger timely interventions in order to try to avoid the disengagement before it takes place. This paper proposes an approach to build the necessary predictive models using students' data that becomes available during a course. The approach was employed in an experimental study to predict the decrease of three different engagement indicators in a MOOC. The results suggest its feasibility with values of area under the curve for different predictors ranging from 0.718 to 0.914.

Smart learning environments (SLEs) have gained considerable momentum in the last 20 years. The term SLE has emerged to encompass a set of recent trends in the field of educational technology, heavily influenced by the growing impact of technologies such as cloud services, mobile devices, and interconnected objects. However, the term SLE has been used inconsistently by the technology-enhanced learning (TEL) community, since different research works employ the adjective "smart" to refer to different aspects of novel learning environments. Previous surveys on SLEs are narrowly focused on specific technologies, or remain at a theoretical level that does not discuss practical implications found in empirical studies. To address this inconsistency, and also to contribute to a common understanding of the SLE concept, this paper presents a systematic literature review (SLR) of papers published between 2000 and 2019 discussing SLEs in empirical studies. Sixty eight papers out of an initial list of 1,341 papers were analyzed to identify: 1) what affordances make a learning environment smart; 2) which technologies are used in SLEs; and 3) in what pedagogical contexts are SLEs used. Considering the limitations of previous surveys, and the inconsistent use of the SLE concept in the TEL community, this paper presents a comprehensive characterization Manuscript sent for review July 30, 2020. Revised month day, year; Accepted month day, year. Date of publication month day, year.

A new architecture called muARTMAP is proposed to impact a category proliferation problem present in Fuzzy ARTMAP. Under a probabilistic setting, it seeks a partition of the input space that optimizes the mutual information with the output space, but allowing some training error, thus avoiding overfitting. It implements an inter-ART reset mechanism that permits handling exceptions correctly, thus using few categories, especially in high dimensionality problems. It compares favorably to Fuzzy ARTMAP and Boosted ARTMAP in several synthetic benchmarks, being more robust to noise than Fuzzy ARTMAP and degrading less as dimensionality increases. Evaluated on a real-world task, the recognition of handwritten characters, it performs comparably to Fuzzy ARTMAP, while generating a much more compact rule set.

With the emergence of MOOCs, there is a growing interest in prediction research. Most existing predictive models do not consider the context for which they are intended, thus resulting in limited impact. Learning design (LD) can provide a contextual understanding for the design of predictive models in collaboration with the instructors, maximizing their potential for supporting learning. This paper presents the findings of a mixed-methods research that explored the potentials emerging from aligning LD and LA during the design of a predictive analytics solution and from involving the instructors in the design process. The context was a past MOOC, where the learner data and the instructors were accessible for posterior analysis and additional data collection. Through a close collaboration with the instructors, the details of the prediction task were identified, including the learning activity to focus on, the target variable to predict, and the practical constraints to consider. Later, two predictive models were built for the prediction task identified: LD-specific model, in which the features were based on the LD and pedagogical intentions, and a generic model, which was based on cumulative features, not informed by the LD. Although the LD-specific predictive model did not outperform the generic one, some features derived from the LD and pedagogical intentions were predictive. The quantity and the power of such features were associated with the degree to which the students acted as guided by the LD and pedagogical intentions. The leading instructor's opinion about the importance of the learning activities in the LD was compared with the results of the feature importance analysis. This comparison helped identify the parts of the LD that need improvement. That is, the results of the LA informed back the LD, where the instructor was a mediator. The implications for improving the LD are discussed.