In an enterprise setting, a major challenge for any data mining operation is managing data streams or feeds, both data and metadata, to ensure a stable and certifiably accurate flow of data. Data feeds in this environment can be complex, numerous and opaque. The management of frequently changing data and metadata presents a considerable challenge. In this paper, we articulate the technical issues involved in the task of managing enterprise data and propose a multi-disciplinary solution, derived from fields such as knowledge engineering and statistics, to understand, standardize, and automate information acquisition and quality management in preparation for enterprise mining.
An online classification algorithm for large scale data streams: iGNGSVM
