A Weighted-Tree Simplicity Algorithm for Similarity Matching of Partial Product Descriptions

Our weighted-tree similarity algorithm matches buyers and sellers in e-Business environments. We use arc-labeled, arc-weighted trees to represent the products (or services) sought/offered by buyers/sellers. Partial product descriptions can be represented via subtrees missing in either or both of the trees. In order to take into account the effect of a missing subtree on the similarity between two trees, our algorithm uses a (complexity or) simplicity measure. Besides tree size (breadth and depth), arc weights are taken into account by our tree simplicity algorithm. This paper formalizes our buyer/seller trees and analyzes the properties of the implemented tree simplicity measure. We discuss how this measure captures business intuitions, give computational results on the simplicity of balanced k-ary trees, and show that they conform to the theoretical analysis.