A hierarchical cluster algorithm based on binary model

It describes the similarity calculation method based on positive attribute distance, cluster evaluation criterion and the clustering process. The criterion mentioned in this alogrithm helps to measure the quality of each clustering level, and with the process of combinating each cluster, the level with the best quality can be seemed as the final cluster, while the cluster number of the level will be the best number. Several experiments on the UCI datasets prove the validity of our algorithm.