Abnormal Data Detection Method of Web Database Based on Improved K-Means Algorithm

A k-means clustering based anomaly detection method for software test data is proposed to enhance the detection ability of software test anomaly data. The distribution document model of abnormal teaching data is established based on portable multi-dimensional control software; Identifying software parameters based on semantic features and extracting feature quantities of software related information; Clustering of abnormal data by feature combination analysis according to feature distribution; Through the fusion of abnormal feature distribution, the joint detection of multi-dimensional features is completed; K-means clustering is used to obtain the optimal data combination and complete data anomaly detection. The experimental results show that the advantages of this method are good performance and accuracy.