|M.Sc Student||Kasif Liat|
|Subject||Enterprise Knowledge Process Modeling|
|Department||Department of Industrial Engineering and Management||Supervisor||Professor Reuven Karni|
Modern organizational IT systems are defined and activated by business process models. These include a significant number of decision activities, such as problems with production capacity or changes in sales orders. They do not, however, specify how decisions are to be taken. We identify these decision points, and provide support for decisionmaking, using a knowledge process based on data mining for discovering significant knowledge in the centralized ERP database. Revealed patterns of behavior can aid the decisionmaker in coming to a conclusion or selecting one of several alternatives. A handbook containing about 230 business logistics process maps has been surveyed. 50 decision nodes have been identified in 40 processes, characterized by decisionmaker, decision type, trigger, circumstances and time window. Most of these decisions are operational; and deal with unplanned or critical situations. They are triggered by the emergence of problematic situations, with little time for deciding. They involve decisionmakers in all the business areas encountered in the handbook. We propose two knowledge procedures. A one-time "preparation", includes node identification, specification of decisionmaking criteria, data mining "target fields", selection of data tables in the ERP database, addition of derived fields, formation of data mining queries and definition of the "decision support database". An ongoing KDD procedure incorporates selection of target fields, generation of a current "decision support database", specification of parameters for "meaningful" rules, execution of data mining, retrieval of complementary information from the ERP database, interpretation of the significant rules in conjunction with this information, decisionmaking and continuation of the business process.