Automatically Identifying Bounds on Semantic Annotations for Bioinformatics Web Service Input Parameters Ravinder Singh, Sean Bechhofer, Khalid Belhajjame, and Suzanne M. Embury School of Computer Science, University of Manchester Oxford Road, Manchester, UK {singhr,seanb,khalidb,sembury}@cs.man.ac.uk Abstract. Semantic annotation of web services is achieved by associat- ing parts of service interface descriptions to concepts defined in a seman- tic model, such as an ontology. Annotating web services with semantic metadata is currently a manual procedure that is conducted by human annotators. In practice, however, service annotation can be both time- consuming and error-prone to implement on real-world web services. We have investigated this problem in the bioinformatics domain, where cer- tain characteristics shared by many web services make them a difficult breed to annotate with semantics. By taking these characteristics into consideration, we present a probing-based technique that identifies up- per and lower semantic bounds for the annotation of web service input parameters. The resulting bounds can serve as a useful starting point for human annotators and could help reduce the required effort in the semantic annotation process. Key words: semantic web services, semantic annotation, bioinformatics 1 Introduction Semantic web services [1] are seen as the next step in the life cycle of web ser- vice technology. The primary goals driving semantic web service research are to automate the discovery, composition, monitoring and execution of web ser- vices. To achieve these goals, web services must first be annotated with semantic metadata. Semantic metadata for web services is provided alongside a semantic model, such as an ontology. Semantic annotations for a web service can then be created by mapping parts of a web service description file (hereafter referred to as WSDL file) to appropriate concepts in an ontology that represents the domain of the web service. Semantic annotation of web services is a task that is primarily conducted by human annotators. Typically, annotators were not involved in the development of the web services they annotate. As a consequence, they face certain hindrances in the annotation process. First, to annotate a web service, a human annotator must be familiar with the functionality that the web service provides. Such infor- mation is obtained by reading any available documentation for the web service