Concept of Semantic Knowledge Base for Data Mining of Business Rules Stanislav Voj´ ıˇ r Department of Information and Knowledge Engineering, Faculty of Informatics and Statistics University of Economics, Prague, Czech Republic, stanislav.vojir@vse.cz Abstract. This paper presents a concept of a semantic knowledge base (in RDF form). This knowledge base will be usable for combination of data mining of association rules and definition of business rules sets. The data mining and domain experts will be able to use it for extension of possibilites of definition not only user-defined business rules, but also business rules generated from association rules gained from data mining tasks. This research is a part of EasyMiner project. Keywords: data mining, association rules, business rules, knowledge base, background knowledge 1 Introduction Let’s consider a business analyst, who is working in a small bank. For example, let’s call him David. David should prepare a classification model for assessing the creditworthiness of clients. He has at his disposal a dataset describing his- tory of loans repaying in last 3 years. Applying data mining algorithms yields a classification model in form of rules. David wants to combine gained classifi- cation model with an older rule set with description of problematic clients. The older rule set was prepared by a domain expert one year ago and was successfully applied yet. Currently, David has a problem: The rule sets are based on different data dictionaries, with another names of attributes with another named groups of values. He would like to have a complex data dictionary for combination of rules from different sources. Today, there is an increasing demand for decision support systems. The limi- tation of quality of each decision support system is quality and complexity of the used knowledge base. Suitable form of a knowledge base is a set of business rules. The advantage of knowledge base in a form of business rules is its modularity. The saved rules can be independently interpreted and evaluated. For combining of rules obtained from multiple resources, it is necessary to define all rules using one shared data dictionary (in business rules terminology, it is called ”terms dictionary”). This paper presents a concept of knowledge base (in RDF form), which is suitable for support of data mining of classification business rules.