The amount of documents available on the Internet makes information retrieval a difficult task. In parallel some efforts towards the Internet personalization allow web pages user-adaptivity. In order to fulfil those objectives current works rely on adaptive information systems and documents summarizing concepts. In this context most systems intend to represent a diversity through a synthesis or a narrative coherence through spatial organisation. We intend to extract a representative collective presenting a thematic coherence according several analytic points of view in order to generate a synthetic document which could be used for data mining. The considered system will be at first designed for the medical field and in particular in the Tissue Microarray technology field.