Context Weighting for Ubiquitous Learning Situation Description: Approach Based on Combination of Weighted Experts' Opinions

Situation awareness is an emerging concept in ubiquitous environments, particularly the learning ones. Situation identification techniques developed in literature aim to infer the user's situation from the detected context. Most of the studies to date in ubiquitous learning (u-learning) field give equal importance for detected context elements to describe learner's situation and therefore allow irrelevant context elements to have as much effect on situation description as relevant ones. Therefore, different weights need to be associated to context elements with reference to their importance to the inference process. In this paper, a new proposal for weighting u-learning context elements is detailed. The solution aims to merge different expert opinions about context elements weighting for situation description. Evidence theory is applied in order to handle uncertain and conflicting expert opinions. Experimental results are given to illustrate the applicability of the proposed solution in selecting characteristic context elements appropriate for each u-learning situation pattern and distinguishing them from others.