Grouping dispersed learners with recommender systems

Wolfgang A Halang and Maytiyanin Komkhao
FernUniversität
Hagen, Germany

Sunantha Sodsee
King Mongkut’s University of Technology North Bangkok
Bangkok, Thailand

Distance learning, in particular e-learning, is characterized by geographically dispersed student populations and only in rare cases do students know each other. The consequence of this impersonality is social isolation, resulting in feelings of loneliness, decreased motivation and, finally, passiveness. To address the needs of geographically dispersed students, several collaborative learning environments were developed, which — in addition to the proper information delivery and communication features within given groups — comprised some forms of group-building tools. Just facilitating contacts with computerized support is, however, insufficient to form effective peer groups whose joint work motivates their members. There is an urgent need to find an effective way to automatically organize learners into highly coherent groups which share common learning interests, and similar personal preferences and dynamic learning behaviours, and also enable individual students to provide capabilities which complement those of their peers. To form cohesive learner groups within large student bodies, the use of recommender systems is proposed as this appears to be quite well suited for the purpose, and also helps to assign students registering late to already existing groups. The basic idea is to bring together people who share common interests and similar attitudes, and whose capabilities are complementary, allowing them to provide mutual help. The special form of consensual recommender systems allows one to characterize student groups. Distance functions which measure the similarity and complementarity of students are the key to successful grouping. Course instructors need to elaborate them according to the educational objectives being pursued.