With the development of Web services technologies, more and more public Web services have been published on the Internet. During the searching and utilizing of these public services, services' textual descriptions (such as introduction and user manual), which are generally expressed in natural language, provide great help for service consumers to locate, understand, and utilize proper Web services. Existing methods for services discovery usually try to obtain such descriptions only from services' WSDL files. However, according to this investigation, lots of Web services do not contain enough textual descriptions in their WSDL files. This paper proposes an approach to enriching textual descriptions for public Web services on the Internet using the information sources outside of WSDL files. Given a Web service, the study collects related Web pages containing its features from the Internet. Then, the enriched descriptions for the service are identified from the Web pages using information retrieval technologies. Experiments conducted on real data indicate that our approach can enrich descriptions for about half of the public services on the Internet effectively. The collected data is publicly available on the Internet.