Elsevier academic partners want to stay at the cutting edge of research and ensure that their work is impactful in society. One way they do it is to have their PhD students work on real-life projects in industrial settings. Elsevier has embarked on a program to support the PhD researchers doing work in applied data science. It gives researchers a chance to work with Elsevier staff and data that are relevant to real life challenges. It also gives Elsevier teams an opportunity to tap into young talent, keep a finger on the pulse of academic research and exercise research and mentoring skills.
This year Elsevier is supporting 12 PhD students at six centres, with the staff acting as industrial sponsors:
But what does it mean to be an industrial PhD supervisor?
We asked George Tsatsaronis, VP Data Science, Research Content Operations, in Amsterdam. George has a PhD in Text Mining and has acted as both an academic and industrial PhD supervisor in his professional roles at TU Dresden, Transinsight GmbH, and Elsevier.
- Why did you take on the role of an industrial PhD supervisor?
Personally, I enjoy doing research and thinking as a researcher. The idea of being able to participate in developing a study from scratch is really satisfying. I love when I’m the first to take something experimental and apply it in a real system. It’s like someone makes a new prototype car and you’re the first to drive it.
- How does the role as an industrial supervisor differ from an academic PhD supervisor?
They’re complimentary roles. The PhD supervisor has the responsibility to ensure that a PhD project is designed correctly. They need to ensure that a student does research in the area that hasn’t already been explored. It needs to be substantial enough for a PhD project, but it also needs borders.
The industrial supervisor has the responsibility of coming up with the best possible application of the outcome of the student’s research. It is a big responsibility. If the application we choose for the algorithm isn’t appropriate, it might show no impact. But that’s because it’s not applied properly, not because of the algorithm [itself]. Normally a person who takes a role of an industrial supervisor is experienced enough to pick which application of the research project will have demonstrable impact.
- What kind of skills/experience do you need as a PhD supervisor? Do you have to have a PhD yourself?
It’s not necessary, but it’s helpful. It’s helpful because you understand the process and structure of a PhD program. Your role as an industrial supervisor is to help students showcase their research and help them graduate. They [PhD students] need publications, generally, so it’s helpful if an industrial supervisor understands what research makes a publishable unit, what quality of work is publishable, what’s been done already in the academic field, what’s considered cutting edge – all of that is useful.
Industrial supervisors also need to be good people managers with a willingness to dive deeply into the latest published work. They need to help students position their research in comparison to other current work and help them shape the project to fill in what is missing. Students can’t graduate just because their work is applicable. It needs to be novel and contribute to the field, so the industrial supervisor needs to be up-to-date with the field.
- What projects are the most appropriate for industrial PhDs?
Ideal projects are the ones on the roadmap for products 2-3 years down the line (Horizon 3). I would not attempt to bring PhD students onboard and co-supervise them to contribute to something I need to deliver this year. With a Horizon 3 timeframe, students can work at their own pace without stress and they can validate their work appropriately and publish it as required.
Projects should also have specific requirements in terms of objectives and success criteria. We can’t ask a PhD student to “improve the Scopus set.” They need a project that is well-defined and specific.
- What value does it bring to Elsevier?
It clearly brings us talent. My team has supervised several PhD students in the recent past. We were responsible for shaping their work in a way that the outcome would be helpful for our industrial applications. Three of those students came to work at Elsevier as NLP scientists. And now we see that they’re engaged into research collaboration and act as industrial supervisors to new PhD students.
If we really do our best to become a global leader in information analysis and data science applications, we need to nurture an environment in Elsevier that is attractive to the best talent. For that reputation to be built and that environment to be created, we need to prove to the top universities and departments that we have interesting content, people and applications. We need to show that our products and services make difference to society. PhD programs are a unique opportunity to do that. Universities have the raw materials and turn to industry partners, like us, to show impact. This allows us to be a leader in transforming and producing the best work out there in applications that are genuinely important to our society.
Noelle Gracy, PhD, is the Head of the Research Collaboration office at Elsevier. The Research Collaboration Unit handles about 70 research collaborations annually, primarily in areas of machine learning, clinical decision support, research metrics and research integrity.
Image credit: Headway from Unsplash
Picture credit: Elsevier