Customized document research by a stigmergic approach using agents and artifacts

Abstract

Document research in a digital corpus can be considered as a browsing process driven by some information needs. Such browses requires the use of traditional information retrieval tools to select relevant documents based on a query. But they can be improved by the use of customization and adaptation mechanisms in order to refine the representation of information needs. Several factors are useful to influence this customization: user profiles, browsing profiles, semantic proximity of documents, recommendations from other similar users, . . . We propose in this article to treat this diversity of influence by a multiagent system interacting with a shared environment representing the users navigation. We follow a stigmergic approach in which the agents implement different customization factors and modify their shared environment to influence the representation of users needs and the browsing. This multiagent system has been implemented using an artifact layer for the environment.