Interest propagation for knowledge extraction and representation.

Abstract

Due to the increasing number of available biomedical data repositories, providing a comprehensive and intuitive access to information is still a demanding task for Information Retrieval systems. In this work we present an interactive data exploration system that retrieves relevant information by propagating the user's interest within a network. The developed techniques have been applied to two different retrieval tasks useful for biomedical research: the prioritization of proteins related to a disease of interest and the search of publications in the literature. The method relies on a network of biomedical entities, scoring of entities of interest by the user, and score propagation. The assessment of the relevance of the retrieved information confirmed a high accuracy of the presented algorithms for both the domains considered.