The descriptiveness of capabilities advertised on service-oriented architectures provides a promising platform for crafting new knowledge. Service mashup has been introduced as an approach for integrating the information provided from multiple web services into one common operational picture. In the future, scale will be a barrier to these types of approaches. With the entry and exit of large numbers of services on the Internet, it will be difficult to find and suggest the most relevant service candidates for new mashups. In this work, we present an efficient syntactical approach for actively discovering web service candidates for service mashups. This approach leverages the message naming characteristics of the developers and of the target service repository to inform search algorithms. Favorable precision results are described based on experimentation executed on an open repository of web service from the Internet.