An Open Platform for Customized Corpus Browsing Using Agents and Artifacts Approach

摘要

Document research in a digital corpus can be considered as a browsing process driven by some information needs. Such browses require 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,... Existing mul-tiagent approaches for document research have proposed an agent model of resources and processes of a given system. We propose in this article a general model for agent-based document research decoupling browsing management and customization or recommendantion decisions. We follow a stigmergic approach in which agents implement different customization factors and modify their shared environment to influence the browsing results. The openness of this architecture is shown by presenting several variants that can be obtained by the dynamic addition of agents or resource artifacts.