ECIR 2020 – Doctorial Consortium report

The Doctoral Consortium (DC) at ECIR 2020 seems set to become a core part of the conference as for the second year running it attracted a good crowd of both submissions and attendees. The DC is a chance for early-stage PhD students in the wider area of Information Retrieval to present their work (often for the first time) to the community outside their home university. One of the benefits is that students whose contributions have been accepted get teamed up with a more senior researcher or academic who serves as their mentor — with the result that both mentor and mentee can learn something!

This year’s DC chairs, Stefan Rüger and Suzan Verberne, did an excellent job in running the show. To start with they worked very effectively with the main conference organisers in getting the DC off the ground in a fully digital way via Zoom. That on its own was most certainly not as easy as it sounds given the short timeframes we are talking about once it became clear that the corona virus was starting to have a much larger impact than it seemed initially. In addition to that, Suzan and Stefan provided very clear guidelines to both mentors and students right from the start so that the mentoring process was not reduced to providing feedback on the DC presentation but instead a communication channel was encouraged between mentors and mentees way ahead of the conference.

The day was structured in blocks with breaks in between so that each students had 20 minutes allocated for presentations and question-answering with two exciting industry talks and some break-out sessions also part of the programme. I counted close to 40 people “attending” the DC and had the impression that students did benefit quite a bit from the format. This was reflected by the individual feedback they provided at the end of the day following a chat among themselves.

Overall I consider this year’s ECIR Doctoral Consortium to have been a very successful event. Part of the reason was the breadth of themes covered by the student presentations ranging from legal search and entity-focused search to reproducibility, sentiment analysis and graph databases (surprisingly though just a pinch of BERT in there). But I should also point out the pitch of the keynote talks by Dyaa Albakour (Signal AI) and Vanessa Murdock (Amazon) reflected on what it means to do research in industry, giving very complementary insights based on their own experiences. Well done and well received by the audience.

What things could be improved? The mingling was certainly missing despite all effort of having discussion rooms etc. A Zoom meeting simply cannot replace the conference setting. There is also the usual problem of deciding on what events to run in parallel. Running the DC in parallel with the workshops and tutorials (as was done this year) means that students presenting in the DC will miss out on some of the activities they might find most beneficial. But then this is not an ECIR-specific problem.

One way of how the iConference (that was run a few weeks earlier) tried to resolve such clashes was to turn the virtual conference into a longer event without parallel activities. I also found that the day was getting a bit long in the end, so perhaps scaling it down a bit might be an option. And obviously, the drinks at the end of a long day are simply not the same in an empty office talking to a screen when compared to what it could have been in the beautiful setting of Lisbon …