recsys 2017, my review

Our biggest conference so far with 620+ atteendes, in one of the most beautiful locations ever, and a program with 46 scientific papers and 12 industry papers, the ACM Recommender Systems conference delivered on its promise with really good content, great organization and an overall open and nice atmosphere.

When it comes to content, some topics are starting to get more and more attention within the community. Tourism, kids and health had their own workshops, and fashion was heavily represented with 4 talks in the main conference that looked at different topics such as deep learning, size recommendation, practical lessons from production systems (by my Zalando colleague Antonino Freno), and outfit recommendations.

With its own fully-booked workshop, a tutorial, and a session in the main conference, deep learning is growing in the conference, and it is doing it while creating some controversy…

untitled collage, 2016 (Humberto Corona)

Beyond content, there are also other topics worth discussing, such as human diversity in the conference (not algorithmic diversity), the industry-academia relationship, and etiquette.

Gender diversity is still being an issue in the conference. We had an all male panel, and also all keynote speakers were men. However, we have a really diverse community of researchers both in industry and academia, and we should definitely have a more balanced gender representation in the lineup when it comes to panels, keynotes and tutorial speakers, so that it represents our community in a more meaningful way.

We, (specially men) need to start looking into ways we can support internal initiatives like the womens’ lunch, we need to encourage our women colleagues (and our gender non confirming ones as well) to submit their work to the conference, we need to encourage them to give keynotes, we need to send emails to the chairs when we see a really good speaker, we need to point out that it is not ok to have only men on the stage, and specially we need to learn to stop complaining (and start asking how can we help) when women decide to have lunch together.

As we grow (we became a two-track conference in 2015 and started to sell-out in 2014) one of the biggest questions we still need to solve is the balance between industry and academia in the conference and our community. It was pointed out (by @agelesschronicl and others) that “Industry attracts 80% of participants, academia presenters have no audience”. However, there is no doubt that industry needs both the research of academics and the academics themselves! One possible solution pointed out by a few attendees is to have topic-based sessions, with a combination of academic and industry presentations. I am sure there are many other possibilities.

A few years ago in 2014, I was presenting at the RecsysTV workshop in on the last day of the conference. On top of it being difficult to get an audience on the last day, I had to compete with two multi-billion dollar sponsoring companies which had announced tours to their offices at the same time my talk was on. As a PhD student presenting in the conference for the first time, it was really hurtful !

This year, we all had had similar experiences. There were at least four different private parties (some more secret than others) and I didn’t get invited to most of them. Just when I thought I had got over my teenage years of not being invited to the cool table, now I don’t even get invited to the geek table! — nevermind, that is ok (but invite me next year please)

To me, the future of the conference is really bright, we are growing and the conferences get more and more interesting, yet we face the same issues our society and industry is also facing and we need to talk and address them.

Finally, and I am delighted to serve as a publicity chair for #recsys2018. If you have ideas on how we can make the social media streams for the conference better, please contact us! Otherwise, see you in Vancouver!