Human-style WWW summarization

After illustrating with the help of a textuality checklist how it transpires that humans can produce so much better summaries than automatic systems, I plead the cause of a summarization system that follows the approach of human summarizers not in the system underground but near the user interface where users actually feel it. I present the overall research strategy and a sketch of the target system that emphasizes some key ideas. First, the preexisting empirical model and computer simulation of professional summarizing is reshaped for its target environment "WWW summarizing in bone marrow transplantation". Second, the domain ontology for knowledge-based human-style processing is explained, together with an empirical procedure for its development. Third, I discuss the interpretation and relevance assessment procedure in some detail. There, the influence of human summarizers is most prominent: agents derived from human summarizers’ strategies are at work. At the time of writing, work on implementation of the target system SummIt-BMT (Summarize It in Bone Marrow Transplantation) has begun.