WSDL Information Selection for Improving WebService Classi cation

Abstract

Currently, the increasing number of available Web Services
(WS) over the Internet has induced the urgency for proposing new ways
for searching and categorizing such software pieces. Normally, WS functionality
is detailed through the WSDL description language, resulting in
a structured document that includes a great variety of features de nition.
One of the WSDL inner features "documentation" is designed to describe
the Web Service functionality, in natural language, which could help
to classify and  nd WS. Nevertheless, the majority of WS lack of that
description. To tackle this problem, this paper presents an analysis of the
WSDL inner feature information that can assist to classify WS, without
any extra data. The experiments carried out on three di erent WSDL
collections showed that only with minimal information is possible to
increase the performance of automatic WS classi cation