Information Management is an important component in the development of viable information sharing environments and systems. Information published in shared environments may be manipulated to generate or derive subsequent information. The qualities of such derived information depend on qualities of the contributing units of information and of the entities that published them. The ability to capture and propagate data about such manipulations through successive generations of information is essential to the calculation of the qualities of information. Such qualitative and pedigree data about the information can provide a basis for making application specific decisions and attributing confidence to those decisions. The cost of propagating pedigree data increases as the volume of information increases through generations. Without loss of generality, most users of an information sharing environment could be satisfied with approximate and quick assessments of qualities and an abbreviated history of related information. However, not all assessments need to be performed in a time critical environment and not all approximate and quick information assessments can satisfy every quality need. In this paper, we present the architecture for an Agent-based Information Management System (AIMS) that provides for mechanisms for establishing quick and approximate qualities and related provenance (to accommodate time-critical decision making, and limited data storage), complemented by a mechanism that supports exhaustive treatment of data provenance, and parallel calculation of precise qualities. We present the novel concept of a Variable Information Chromosome per unit of information that propagates only the most representative provenance across generations. The multi-agent framework provides extensive support for provenance related services.