A/B/C-TEACH adopts a holistic approach of the fundamental channels from which the educational process is conveyed, combining cognitive, affective and social information of the peers’ behaviour and interactions. The architectural structure of the A/B/C-TEACH is depicted in the figure below. From the latter, the dynamic flow of information between the different structural modules and participants in the A/B/C-TEACH framework is evident.
The interconnected elements of the figure include:
Users: M Teachers and N Learners are associated with J courses interacting both F2F and through the LMS under the b/c-learning concepts [1,2].
LMS Moodle-Metrics and Collaborative Interactions: LMS Moodle is intentionally built on pedagogical strategies (e.g., behaviorism, cognitivism, constructivism, connectivism), allowing management of user data, usability issues and exhibiting adaptation capabilities . In this vein, a series of metrics regarding the interaction and collaboration amongst users are available. In particular, in a recent work , 110 metrics were used (e.g., wiki, blog, forum, chat, quiz, edit) forming interaction qualities referring to view, addition, alteration, action and contributed to the estimation of QoI. In the work of Hadjileontiadou et al. , 10 collaborative interactions (e.g., proposals, contra-proposals, questions, turn-taking balance) were used for the estimation of the QoC.
Affective Module: The rest 1/3 of the analysis dataset will be covered by the data from the learners’ affective module, supporting a-learning . This module will avoid the typical path of evoking questionnaires to measure the learners’ affective experience — such as how much pleasure, frustration, or interest they felt during the learning processes, and evaluating the motivational characteristics of an instructor’s delivery. On the contrary, by using affective computing cutting-edge technology (such as EEG EMOTIV EPOC/INSIGHT, Microsoft Kinect 2, portable functional near-infrared (fNIR) spectroscopy) it would provide real-time data, related to the users’ ASs.
Modelling: Considering the data from the previous modules, the modelling one will use innovative techniques from the fields of fuzzy set theory [4,5], neuro-fuzzy modelling , intuitionistic fuzzy systems , dynamic nonlinear analysis , and affective computing . In particular, the experts’ knowledge will be codified in the form of fuzzy rule bases (i.e., a set of fuzzy IF/THEN rules), also modelling the experts’ hesitancy and the nonlinear mechanisms in the production of collaborative data, in order to produce hybrid models that could be flexible and adaptable to the characteristics of each user. Moreover, by adopting Lang’s emotion space  combined with advanced signal processing techniques , emotion recognition processes of the AS data will take place.
Features: The main features that will be outputted from the constructed models include users’ QoI, QoC and AS (learners only).
Feedback: Having the aforementioned estimated features per user, the construction of personalised feedback will be fired. The latter will be focused upon initiating metacognitive processes, helping the users to become more aware of their interaction, collaboration and affect — building a kind of ‘interactive/collaborative/affective mirror’ in which the learners are encouraged to reflect upon how their interaction/collaboration behaviour and state are influencing their learning experience. For example, emotional awareness, in oneself and in others, is considered to be a learnable skill of emotional intelligence. Being aware of one’s state, such as frustration, can be instrumental in helping deal with that state productively. Moreover, enriched feedback regarding more global findings will be provided to the HEI’s policy stakeholders towards iLMS.