The Tree-to-Tree Correction Problem

Abstract

The tree-to-tree correctmn problem Is to determine, for two labeled ordered trees T and T', the distance from T to T' as measured by the mlmmum cost sequence of edit operaUons needed to transform T into T' The edit operations investigated allow changing one node of a tree into another node, deleting one node from a tree, or inserting a node into a tree An algorithm Is presented which solves this problem m time O(V* V'*LZ* L'2), where V and V' are the numbers of nodes respectively of T and T', and L and L' are the maximum depths respectively of T and T' Possible apphcatmns are to the problems of measuring the similarity between trees, automatic error recovery and correction for programming languages, and determining the largest common substructure of two trees KEY WORDS AND PHRASES tree correction, tree modification, tree similarity CR CATEGORIES 3 79, 4A2, 4 22, 5 23, 5 25