A structural classifier to automatically identify form classes

Abstract

This article deals with the description of a new classifier for an automatic form class identification system. This new structural classifier is based on a tree comparisons. The high level information used by this classifier is presented in the article. A module first extracts the form content. The form content organisation is described in a hierarchical way modelled by a tree. This tree corresponds to the input features of the structural classifier. Experimental results are presented and several strategies of combined uses of this structural classifier with other classical classifiers are suggested in order to enhance the results.