A documentation infrastructure for the management of data mining projects

Effective project management is a key factor for successful Knowledge Discovery in Databases (KDD) projects. The systematic documentation of previous knowledge, experiments, data and results is a helpful means of keeping track of the project current status. Despite its value, documentation is most often perceived as an overhead. We propose a documentation infrastructure composed of a documentation model and a supporting environment that allows the capture, storage and retrieval of KDD process-related information and artifacts. The paper describes this infrastructure, and reports preliminary experiences on its use. Preliminary results reveal generalized satisfaction with regard to infrastructure expressiveness and functionality, and highlight the contributions of the documentation produced for improving project management, project execution and team communication. The role of documentation in learning and reuse was also identified.