Context-aware systems assist people?s everyday tasks by adapting their behavior based on context information gathered from instrumented environments. Context is any relevant information describing entities in a user-computer interaction. A pertinent issue to developing such systems is how to represent context. High-level context models prevent the development of ad hoc and limited schemes for context management. Moreover, the more formal a context model is, the better is the ability for reasoning about context. This paper presents our work toward a domain-independent ontology-based context model, which provides a set of general classes, properties, and relations so lower ontologies can import them for particular domains. We show how this context model can be extended in a ubiquitous e-learning scenario on campus. Built upon our model, we also present a configurable service infrastructure for the prototyping of semantic context-aware systems.