Text Mining for Domain Structure Analysis in a Training System for Electrical Procedures

Learning environments themselves constitute a source of knowledge to understand and improve educational settings. We have developed some training systems where instructional content consists of separated electrical procedures stored in text documents. Since, the adequate characterization of the domain is a key factor in learning, we conducted a text mining study to understand relationships between topics, tasks and procedures in the training systems. In turn, this knowledge can help to provide students with an adaptive training, to help instructors to plan courses and to improve the training systems. We rely on a machine learning approach applying bag of words model and a clustering algorithm. Results on mining partial instructional content are encouraging since they show some useful relationships. Here, general proposal and preliminary results are presented.