Knowledge discovering for document classification using tree matching in Texpros

This dissertation describes a knowledge-based system for classifying documents based upon the layout structure and conceptual information extracted from the content of the document. The spatial elements in a document are laid out in rectangular blocks which are represented by nodes in an ordered labelled tree, called the layout structure tree (L-S Tree). Each leaf node of a L-S Tree points to its corresponding block content. A knowledge Acquisition Tool (KAT) is devised to create a Document Sample Tree from L-S Tree, in which each of its leaves contains a node content conceptually describing its corresponding block content. Then, applying generalization rules, the KAT performs the inductive learning from Document Sample Trees of a type and generates fewer number of Document Type Trees to represent its type. A testing document is classified if a Document Type Tree is discovered as a substructure of the L-S Tree of the testing document; and then the exact format of the testing document can be found by matching the L-S Tree with the Document Sample Trees of the classified document type. The Document Sample Trees and Document Type Trees are called Structural Knowledge Base (SKB). The tree discovering and matching processes involve computing the edit distance and the degree of conceptual closeness between the SKB trees and the L-S Tree of a testing document by using pattern matching and discovering toolkits. Our experimental results demonstrate that many office documents can be classified correctly using the proposed approach