ANALYSIS OF LEARNER PROFILE GENERATION ALGORITHMS

摘要

With the rapid growth of computer and Internet technologies, e-learning has become a major trend in the computer assisted teaching and learning fields. By observing how learners behave during their online self-study, educators are then capable of comparing, evaluating, and profiling individual learners learning processes and thus making suggestions to learners with similar characteristics in the same context. Learner profile generation can be achieved by various methods like sequential data mining algorithms where computer logs are analyzed to profile learners in terms of their learning tactic use and motivation in a web-based learning environment. Basic steps involved are preprocessing, pattern discovery and pattern analysis- evaluation. Another approach is that of Fuzzy Cognitive Map (FCM) tool which is a soft computing tool and the reason, which leads to FCM approach, is mainly the observation of uncertainty in learners profile description. Therefore, classes in any classification of learners profile are considered as fuzzy sets and are represented as vertices of a Fuzzy Cognitive Map. It is based on Kolbs learning model which is widely accepted technique. A third approach is that of using genetic algorithm based on adaptive learning for fulfilling multiple constraints to determine the learning scheme which best suits a learner. Adaptability can be provided at different levels according to the context of the learners. For constraint satisfaction problems in which multiple alternative paths have to be explored, genetic algorithm based approach is best suited. Literature survey done on the above approaches shows that a lot of work is being carried out in the area of learner profile generation and understanding of the various approaches. In this report exhaustive study on various methods like genetic based algorithms, adaptive learning based on Kolbs learning cycle using FCM tool and sequential pattern analysis is presented.