A frame work for search forms classification

The deep web pages provide highly relevant quality content and represent a large sector of online information sources, yet general purpose search engines do not find deep web pages due to difficulties of identifying them. In this paper, we present a frame work to identify search forms from non search forms using a small number of HTML input elements extracted from user input HTML forms and a few keywords. It utilizes pre-query technique and post-query technique in a hierarchical manner. Decision trees and multi layer artificial neural networks were used to obtain the classification rates over 91% to classify search forms and non search forms. In the new frame work, it is proposed to use post query technique additionally at the last step to distinguish suite search forms form deep search forms.