Integrating knowledge gained from data mining with pedagogical knowledge

Discovering knowledge from raw data is one of the goals of data mining. Yet, it is not always clear how this knowledge is used in educational computing systems and how exactly it is integrated with other knowledge like the pedagogy used. We present a case study where the use of the data mining results was initially described, to a large degree, at the implementation level, thus largely ignoring the nature of the different kinds of knowledge involved. Based on Clancey's heuristic classification model [7], the description is raised to a conceptual level, the knowledge level. This results in an explicit and well-defined integration of knowledge discovered with data mining techniques, pedagogical knowledge and linguistic knowledge. Such a knowledge-level description leads to an improved understanding of the system and its effects on the learners.