Abstract
An e-learning data warehouse is nourished by view definitions built upon heterogeneous, distributed, and autonomous e-learning information sources. It is frequently analysed and/or mined by e-learning actors such as educators and learners for decision making various reasons and purposes. The above e-learning view definitions, which represent considerable educational information resources, can become undefined when the underlying e-learning information sources change their schemas accordingly to their autonomy characteristics. This obviously decreases e-learning resources availability and consequently affects analysis and mining efficiency. In this paper, we propose to study the issues of using agent's based architecture to achieve the e-learning data warehouse maintenance under schema changes by automatically restoring affected view definitions. This is ensured by offering a solution that guarantee view definitions' flexibility which implicitly optimizes e-learning resources availability by automatically finding replacements, using in a first step static agents and in second step mobile ones, for affected components belonging to view definitions and representing critical information necessary for analysis and/or mining.