Indexing Tree and Subtree by Using a Structure Network

Encoding the clustering result to represent the structure feature of a tree ‹ Encoding each cluster as following field ‹Link all cluster fields to make a long numeric array that can represent the structure of a tree Describing the trees' super‐sub relationships with a network, and using a numeric method to access it fast. PURPOSE ‡ Structure Network: A network describing the super‐sub relationship of trees with the tree structures as the nodes. ‡ Find out super & sub trees for a given tree ‹ Super‐tree: Adding an edge ‹ Sub‐tree: Removing a leaf edge ‡ The structures of super & sub trees of a tree are corresponding to the results of clustering its vertices ‹ Clustering vertices by using the histograms of graph distances By using this network the trees' super‐sub relationships can be represented. ‡ Each vertex can be labeled with its graph distance histogram ‡ The labeled results are corresponding to the labeled result in a traditional matching method [1] The clustering result includes all the structure information of a tree ‡ Comparing the isomorphic ability with the traditional matching method [1]