Modeling Documents for Structure Recognition Using Generalized N-Grams

In this paper we present and discuss a novel approach to modeling logical structures of documents, based on a statistical representation of patterns in a document class. An efficient and error tolerant recognition heuristics adapted to the model is proposed. The statistical approach permits easily automated and incremental learning of the model. The approach has been partially evaluated on a prototype. A discussion of the results achieved by the prototype is finally made. 1 Introduction The need for strongly structured documents has been recognized for a long time and their importance increases with the development of new software applications. Several approaches and systems have been proposed to address this still open issue [4, 7, 8, 10, 16, 13]. It is commonly admitted that documents can be represented by their hierarchically organized physical and logical structures. The recognition of the logical structure roughly consists in mapping a logical tree on a physical one. The protot..