Development and Evaluation of a Math mCSCL Platform with Aggregated Content-Based Filtering and Guessing Detection

Abstract

This study reported the development and evaluation of an intelligent math mobile computer-supported collaborative learning (mCSCL) system for solving fractions. The software included a recommender system and a guessing detector. The aggregated model of a content-based filtering (CBF) algorithm for a group was used to build the recommendation function of the software. The dice coefficient was utilized to determine the dissimilarity of the selected game settings vis-à-vis the problem space PPP. Meanwhile, the guessing detector comprised the Rasch model and computational fluency (CF). CF, in turn, had two components: response time and accuracy. All three indicators had to be flagged as true to classify a student as guessing. Fifty-five Grade 5 students from the elementary departments of four universities in Manila participated in the study. Results showed that the software was able to detect the guessing behavior of each student and provided individualized feedback to those who exhibited guessing. The software also generated recommendations (i.e., game settings), confirming the effectiveness of the CBF and dice coefficient. Content analysis revealed that the software received favorable remarks from the students, implying its relevance to their mathematics learning. However, despite these capabilities, students still exhibited under- and over-practice of certain skills. Limitations of the software were also discussed.