Finding approximate patterns in undirected acyclic graphs

摘要

We consider an approximate pattern matching problem for undirected acyclic graphs. Specifically, let P be a pattern graph, D a data graph and t an integer. We present an algorithm to locate a subgraph in D whose distance from P is at most t. The distance measure used here is the degree-2 metric published previously. The time complexity of our algorithm is O(NPNDddlogd) where NP and ND are the number of nodes in P and D, respectively; d=min{dP,dD}; dP and dD are the maximum degree of P and D, respectively. Central to our algorithm is a procedure for finding approximate patterns in rooted unordered trees and freely allowing cuts. We discuss two applications of the algorithms in chemical information search and website management on the Internet.