Prior to the access to decentralized resources like web services and shared files in peer-to-peer networks, the user needs to be provided with accurate information about these resources. While some of them can be specified impartially, other descriptions might be biased by individual preferences or subjective utility, like quality ratings or content synopsizes. Unfortunately, the assessments of distributed resources usually either solely reflect the requirements, opinion and preferences of certain resource providers, or they consist of quite general recommendations obtained from voting mechanisms and interest building. In contrast to these approaches, we propose an agent-based framework for the distributed assignment and social weighting of rich, multidimensional resource descriptions from the possibly conflicting opinions of communicating agents, which compete in the assertion of individual descriptions.