Proactive Documents--A New Paradigm to Access Document Information from Various Contexts

In this paper we introduce proactive documents (PD) and proactive document agents (PDA) as a new paradigm to access document information from various contexts: browsing, other services and other agents. At large, the PD(A) system architecture is driven by one basic assumption about the nature of documents: natural language documents reflect the knowledge of the authors about a specific topic. PDA are intelligent agents, equipped with profound natural language processing (NLP) and information extraction (IE) algorithms. They create a complete representation of the structural and linguistic knowledge about the processed document. Depending on the context of usage, PD can present this knowledge in different ways. In a browsing- context, PD provide a text based browsing interface, providing typed, contextual links to other documents. If accessed by other services/agents PD are able to explicate their knowledge in a formal way, e.g. as an OWL- lite document. Thus, the information can easily be accessed and further processed by the requesting agents or services. Furthermore, we introduce a complete PD(A) architecture. It allows several PDA to communicate with other agents and to consolidate and evolve their knowledge, cluster documents into information items and create typed links between information items. Finally, we outline the interna of PDA: a mathematical model for information value calculation for document constituents.