Abstract of /TIGKI/INFOSYS/FKI/FKI-243-01

Document-Name:  fki-243-01.ps.gz
Title:		Towards a Multiagent System for Competitive Website Ratings
Authors:	Matthias Nickles 
Revision-Date:	February 2001
Category:	Technical Report (Forschungsberichte Künstliche Intelligenz)
Abstract:	Websites usually appear to the web surfer like isolated islands: 
                Their self-description and their hyperlinks linking to other 
                sites reflect solely the requirements, opinion and preferences 
                of the respective site owner, there is no explicit relation to 
                the information, service and product offers of competing sites, 
                and the assessment of the site by other users is unknown as well. 
                In contrast to usual recommendation systems, this paper outlines 
                a multiagent system which derives multidimensional website ratings 
                from the possibly conflicting opinions of interacting rating 
                agents witch compete in the assertion of their individual ratings 
                against others to provide a ``socially enhanced''solution for 
                this well-known problem. The rating agents act as representatives
                for individual web users, interest groups, other web sites, 
                private or public organizations, and represent their ratings in 
                an open discussion forum which is attached to the rated web site. 
                This forum is continuously observed by a rating instance which 
                computes rich, social weighted general ratings of different
                abstraction level from the forum communications. In contrast to 
                the results of usual majority voting based recommendation 
                systems, these general ratings take into account the social 
                structures like norms and roles which emerge from the agents 
                communication process. In addition to the presentation of general
                ratings to the web surfer, the ratings and the social structures 
                can also be queried by the rating agents themselves to improve 
                their rating capabilities.
Keywords:       Multidimensional rating, Collaborative filtering, Semantic web, 
                Multiagent systems, Agents, Socionics
SIZE:		25 pages
Language:	English
ISSN:		0941-6358
Copyright:      The ``Forschungsberichte Künstliche Intelligenz''
                series includes primarily preliminary publications,
                specialized partial results, and supplementary
                material. In the interest of a subsequent final
                publication these reports should not be copied. All
                rights and the responsibility for the contents of the
                report are with the authors, who would appreciate
                critical comments.

Matthias Nickles