Grigorios Tsoumakas, assistant professor at Aristotle University of Thessaloniki, Thessaloniki, Greece


In this talk I will discuss our winning approach at the 2013 large-scale online biomedical semantic indexing task of the BioASQ European Project, where we consistently outperformed, among other contestants such as the National Library of Medicine's Medical Text Indexer system, which supports PubMed curators in indexing articles. I will discuss the lessons learned from our participation in this challenge and present a variety of document representations and learning approaches that we have experimented with, pointing out what has worked well and what not. I will also mention our latest developments as well as first results from the 2014 version of the BioASQ biomedical semantic indexing challenge.