The actual implementation of semantic-based mechanisms for service retrieval has been restricted, given the resource-intensive procedure involved in the formal specification of services, which generally comprises associating semantic annotations to their documentation sources. Typically, developer performs such a procedure by hand, requiring specialized knowledge on models for semantic description of services (e.g. OWL-S, WSMO, SAWSDL), as well as formal specifications of knowledge. Thus, this semantic-based service description procedure turns out to be a cumbersome and error-prone task. This paper introduces a proposal for
service annotation, based on processing web service documentation for extracting information regarding its offered capabilities. By uncovering the hidden semantic structure of such information through statistical analysis techniques, we are able to associate meaningful annotations to the services operations/resources, while grouping those operations into non-exclusive semantic related categories. This research paper belongs to the TelComp 2.0 project, which Colciencas and University of Cauca founded in cooperation.