Eliminating Anomalies in Learner Modeling Using Two-Partial Learner Model

摘要

Sometimes, gathered information from tracking learner interactions is not precise. In fact, existing e-learning systems only know the number of previewed pages precisely. They can deduce that how much the learner is mastered using some learner's characteristics such as time of doing exercises and the number of mistakes. But these deductions may be not precise. Because the time of doing exercises depends on learner's emotional states and environmental conditions. Thus we have introduced a new concept as two-partial learner model. Our proposed model divides learner model into two parts: Permanent Learner Model (PLM) and Temporary Learner Model (TLM). In the two-partial learner model, system's deductions are placed in the TLM at first. Then, system should validate accuracy of these deductions. Valid deductions are used in the updating of PLM for making them usable in other sessions. Otherwise, they should be ignored. Two-partial learner model is suitable for example-based educational systems. Because these systems are making deductions about the learner based on his/her interactions with the system.