The description of public resources such as web site contents, web services or data files in open peer-to-peer networks using some formal framework like RDF usually reflects solely the subjective requirements, opinion and preferences of the resource provider. In some sense, such resource descriptions appear "antisocial" as they do not reflect the social impact of the respective resource and therefore might not provide impartial, reliable assessments. E.g., commercial web sites do not contain any relationship to the information, service and product offers of competing sites, and the assessment, of the site by customers, experts or competitors is unknown to users and information agents also. We introduce an open multiagent system framework which derives multidimensional resource descriptions from the possibly conflicting opinions of interacting description agents, which act as representatives for individual, organizational or institutional clients, and compete in the, assertion of individual opinions against others to provide a "socially enhanced" solution for this problem. In contrast to the results of majority voting based recommender systems, the obtained social resource descriptions reflect social structures such as norms and roles Which emerge from communication processes.
|Title of host publication||Cooperative Information Agents Vii, Proceedings|
|Editors||M Klusch, A Omicini, S Ossowski, H Laamanen|
|Number of pages||16|
|Publication status||Published - 2003|
|Name||Lecture Notes in Artificial Intelligence|