Model-based design methods, and model-based architectures, have gained adoption in authoring applications for the WWW. This is further reinforced by the increasing visibility of the Semantic Web, where models are intrinsic, described as metadata associated with the data made available to users and applications. Several proposals attempt to leverage this additional information to improve search functionalities, by incorporating semantic similarity (or proximity) measures into the search mechanism. In this paper we show how the availability of a semantic similarity evaluation engine can be used to enhance several functionalities of Web-based applications. In particular we will show how such an infra-structure can be used to detect and suggest new relation instances, as well as propose an inferred ordering for the presentation of related information that reflects the semantic closeness of the corresponding information. The proposed engine is based on a hybrid spread activation algorithm applied to the concept instances graph.