Intelligent document retrieval from heterogeneous cardiology literature sources: use of an automated notification robot learning from his user

Information management becomes harder with the increasing flood of information. The user needs systems which take a preselection of the information provided and which report periodically to him on their own. The described project retrieves its content from the Pubmed Knowledge-Base at the NCBl and diverse internal datasources of an in-house document-management solution and links to the corresponding (online) documents. Only for Medline a similar approach is performed by services like http://www.pubcrawler.ie/ and http://www. biomail.org. The special feature of the here described work is the learning/grading function of the software, which allows the semi-automatic improvement of the user (search) profiles. Some of the basic components of this project are shared as open-source with partners.