<![CDATA[Web mining in searching based on keywords by automatic clustering is a document searching method by classifying documents based on its keyword. Following is the clustering by centroid linkage hierarchical method (CLHM) to the number of keywords from each document. In clustering, initialization is commonly required for the number of cluster to be formed first, however, in some clustering cases, the user cannot determine how many clusters can be built. Therefore, on this paper, the Valley tracing method is applied as a constraint which identifies variants movement from each cluster formation step and also analyzes its pattern to form automatic clustering. Document data used are from text mining process on documents. Based on 424 documents, this research shows that clustering method using CLHM algorithm can be generally used to classifying documents with exact number automatically.]]>
