SmILE Semantic Information Systems and Language Engineering Group

Reading guide: This is a report on the Elsevier Grand Challenge final. For a general information on the event, read on the whole post. For a shameless self-promotion of us, the CORAAL team, and a couple of enthusiastic shouts, you can directly jump here... ;)

On Tuesday, April 21th me and Siggi were in New Orleans, where I was presenting the latest development of our CORAAL prototype within the Elsevier Grand Challenge final. CORAAL is a joint work of me, Tudor and Siggi, and it essentially presents a knowledge-based search engine currently deployed on biomedical literature data. For a comprehensive, but still quite concise overview of the system, you can check our recent paper (to appear in the proceedings of the 12th Conference on Artificial Intelligence in Medicine). Before going into the details on the final, I have to emphasise the whole thing would not be possible without the great work and much appreciated support of Tudor, Siggi, and also Ioana, who developed an initial version of the user interface.

As you can see, the day of the final was beautiful:

Also, there was a smell of adventure in the air:

Yet the robotic dinosaur show was not the most adventurous and exciting thing to happen that day in the city of jazz...

The Elsevier Grand Challenge final was organised in two sessions:

1. morning at the Experimental Biology'09 conference - presentations to the judges and attendees of the conference
2. afternoon in the Marriott hotel meeting rooms - live webinar for the virtual audience registered online

The morning session was particularly exciting for those who could participate live. We and the other finalist teams' representatives were presenting the current state of our tools and one could clearly tell a difference from what we saw at the semifinal. Although already the semifinal presentations were astonishingly good, now one could see a lot of additional improvements. These were sometimes quite substantial, and not only in case of CORAAL (which has been reworked almost from scratch regarding both back-end and user interface). The changes were driven mostly by the needs of the potential user base (i.e., life science researchers and practitioners) and it was great to see how all the teams are able to reflect the community demand in such a short time-frame.

There were a lot of tough questions coming from the judges, and sometimes non-trivial efforts were required to answer them, as one can see below.

All the questions were going to the point, though, and indicated the deep insight into the work of the contestants the judges had gained within the challenge. Moreover, the finalists usually were aware of the potential problems raised by the questions and proved that they know how to remedy them within a future development of their tools. In the afternoon webinar session the finalists recalled the essence of the delivered tools and coped with one more round of enquiries from the judges and world-wide audience present on-line.

I believe no one can deny that we, the Elsevier Grand Challenge finalists, have exhibited outstanding skills and drive no matter whether we eventually won a prize or not. As Ed Hovy, the chair of the challenge judges, repeatedly said, everybody who was there was already a winner. There had to be the two absolute winners, nonetheless, and the judging committee had to undergo the painful and difficult process of their selection...

At this point We can proudly announce that we have become one of the two winners of the Elsevier Grand Challenge, ending up at the second place (and getting the related prize money of $15,000)! Here you can see the moment of the final relief after anxious waiting for the eventual announcement, as Herman van Campenhout, the Elsevier's CEO for Science and Technology, awards me with the prize (the incredibly efficient and devoted challenge organisers Noelle Gracy and Anita de Waard - from left to right - are standing in the back):

Of course, it would have been nice to get the first prize. However, the Reflect team lead by Sean O'Donoghue has delivered a truly mature tool addressing a clearly defined and important problem. Moreover, they have already attracted a significant user base when developing Reflect before and within the Elsevier Grand Challenge. For these reasons, they fully deserve the first place. We started to realise our vision of convenient search for knowledge scattered across large amounts of scientific publications only recently, essentially after we made it to the challenge semifinal. Thus the second place is a tremendous success for us. And this is not (only ;) ) because of the money or publicity related to the prize, but most primarily because it means that the first steps towards realisation of our daring visions are already appreciated by the leaders in the field and have a strong potential for real users.

No matter how far we have got, this has only been a start, like for the ship departing towards a sunny morning on the verge of my leave from New Orleans...

Now we plan to take CORAAL to the next level and make it available and applicable for users not only in the biomedicine field, but in any field that requires knowledge-based search in vast amounts of unstructured textual data. This will be done in collaboration with the increasing number of groups interested in our work. Hopefully we will also join our forces with Elsevier and the other challenge participants along the plans that emerged from the fruitful discussions we have had so far. Feel free to let us know if you also want to be involved in this exciting endeavour!