A method and apparatus for point-to-point clustering are disclosed. The stepwise clustering method includes a step of comparing distances from any one of the elements of data to a cluster formed by trees of other elements, and if the minimum value of the distances is not less than a predetermined value, And forming a new cluster with the element of interest by inserting it directly next to a particular cluster, and if the minimum value of the distance is less than a predetermined value, moving the element of interest below the first cluster with a minimum distance, And inserting the element of interest into a sibling node of an element that is a descendant of a particular element. Therefore, it is possible to effectively cluster multiple related data without prior information on the number of clusters, and efficient clustering is possible even in the case of real-time updated data or large-capacity data.
展开▼