Using ANFIS to Efficiently Model Skills and Beliefs in Computer-mediated Collaboration

An approach in modeling collaborative and metacognitive data is presented in this paper. The proposed scheme, namely Collaboration/ Metacognition–Adaptive Network-based Fuzzy Inference System (C/M-ANFIS), uses neurofuzzy structure to adaptively infer on the relation between the above data in a meaningful way. More specifically, the collaborative and metacognitive data refer to the participant’s collaborative skills and his/her beliefs on the quality of his/her collaboration respectively, during sessions of collaboration. The C/M-ANFIS allows the intense analysis of these empirical data facilitating a microgenetic look at how change in collaborative and metacognitive activity occurs across the sessions of collaboration. Moreover, through training procedures, the C/M-ANFIS model manages to estimate the processes that give rise to this change. Furthermore, based on the estimated relationship, the model may predict forthcoming values of a feedback indicator (quality of collaborative activity). This information, combined with coaching messages, may be presented to the users as an enhanced feedback.