Scholar Agent: Auction Mechanisms for Efficient Publication Selection on Congresses and Journals

• Authors: Josep Lluis de la Rosa;Maria de los Llanos Tena
• Venue: Proceedings of the 2007 conference on Artificial Intelligence Research and Development
• Year: 2007

Congresses and journals (CJ) can be like displays which can be used as a publishing medium of scholarly works when space is a scarce resource, and it is desirable to expose many papers, posters and communications (adverts of scientific work) to as wide a scientific audience as possible who will eventually cite them. Although the efficiency of such publishing systems can be improved if the CJ are aware of the identity and interests of the audience, this knowledge is difficult to acquire when users are not previously identified or inscribed in the CJ. To this end, we present Scholar Agent, an intelligent public agent, which helps CJ to select and display papers, posters and communications in response to scientists inscribed to particular sessions or tracks in the audience. Here, scientists are identified and their CJ review history tracked. Within Scholar Agent we have designed an agent system that utilises a new currency and an auction-based marketplace to efficiently select papers for the display. We show, by means of an empirical evaluation, that the performance of this auction-based mechanism when used with bidding strategy, efficiently selects the best paper in response to the audience presence (inscribed). This may have utility both presential or on-line CJ. We show how our scholarly publishing agents (Scholar Agents) will behave accordingly the private value of the paper by the scientist and how to link this initial private value with the own private value of the agents that are bench-marked with two other commonly applied selection methods for displaying papers in CJ.