Context Aware Computing is a paradigm that allows applications to find and take profit from information that surrounds them at a given time, such as location, nearby resources, available infra-structure, user preferences, among others. Research in this area focuses on mobile or ubiquitous computing. However, recent research indicates that several aspects of this area are not limited to mobility and ubiquity. They can also profit from systems connected to other areas. Comprehensive multimedia/hypermedia systems must support all the phases of a document treatment. The treatment must encompass authoring, as well as storage and final presentation to the reader (end user). Particularly in this type of system, it is desirable to have tools that adapt the document, that is, tools that apply transformations to the hyper-document, making it adequate to the visualization. Hypermedia systems with adaptation strategies are an example of context-aware applications. The goal of this dissertation is to create a new architecture for context management: one architecture developed and built as an independent structure that may be reused by different systems, particularly by hypermedia systems. This architecture is based on a model of context information representation and on a recyclable framework that can be incorporated to several implementations.