Text clustering addresses in structuring the large set hypertext browsed by the user. Feature clustering is an important strategy applied which focused on reducing dimensionality of feature vectors for text classification. In the Existing System, presented a fully automated technique to generate wrappers for extracting search result records from result pages dynamically generated by search engines. Our technique utilizes both the visual content features on the result page as displayed on a browser and the HTML tag structures of the HTML source file of the result page. This differentiates our technique from other competing techniques for similar applications. Our experimental results indicate that our technique can achieve high extraction accuracy. But it has It has low accuracy results and less efficiency. We propose a self-constructing feature clustering algorithm for feature clustering. The cluster is classified by employing the aggregate function such as mean and deviation. The extracted feature, corresponding to a cluster, is a weighted combination of the words contained in the cluster.