Recognition Algorithms for Structured Documents with Variable Content

The paper deals with a wide class of structured documents that cannot be described using one or several models based on associations between the document fields and geometric elements. A formal model of such documents is described that is based on the concept of a multiset. Examples of structured documents of this class are given and a technique for the construction of models of structured documents is proposed. This technique is illustrated using an implementation of an automated document management system. Implemented algorithms for detecting document fields are described, and implementation problems are discussed.__________Translated from Programmirovanie, Vol. 31, No. 4, 2005.Original Russian Text Copyright © 2005 by Slavin.