Rule based design using clustering for knowledge acquisition

Data analysis can be done by expert system decisions on system status according to system input and output data. For the purpose of data analysis, there is often a need to classify data or to find regularities therein. The results of the regularity search can be expressed by the IF-THEN production rules. The use of different approaches – with clustering algorithms, neural networks – makes it possible to obtain rules that characterise data. Knowledge acquisition in this paper is the process of extracting knowledge from numerical data in form of rules. Rules acquisition in this context is based on clustering methods. With the help of the K-means clustering algorithm, rules are derived from trained neural networks. The rule-making methodology is demonstrated on a sample basis of IRIS data. The effectiveness of the obtained rules is evaluated.