Mining Uncommon Information from Inconsistent Samples Based on Support Vector Machine

In current researches of knowledge discovery, inconsistent examples in a decision table are not be analyzed. It is just the place that contradictions would hide interesting and valuable information. An algorithm based on the support vector machine is proposed to mine kinds of information which hide in inconsistent examples, i.e., to decide whether inconsistency is caused by mistake, the error between a computed or measured value and a true or theoretically correct value, or missing attributes. Some methods and algorithms which eliminate the inconsistency are presented.