Searching the web proved to be a critical task in which both user satisfaction and speed requirements must be satisfied. Nowadays search engines provide amazing capabilities of searching resources on the web though they are still based on text indexing and seldom exploit semantics of resource content. Therefore they sometimes fail on the identification of query context, providing results that are not relevant with respect to user needs; this is especially true for non-trained users. The Semantic Web specifically addresses such an issue by providing means to define resources and query semantics using ontologies and semantic annotations. In this paper we propose a concept-based search paradigm for document retrieval. We define a new representation of the involved information space, introducing the notion of Conceptual Spectrum to identify the global topic landscape of web resources and queries. We exploit such a representation to design and implement a prototypical version of a concept-based search engine. We compare the proposed approach to a traditional keyword-based engine, on a specific domain. Results are promising and show the approach feasibility.