Doctoral Research today is increasingly shaped by collaborations beyond the university setting. Within the MSCA Doctoral Network funded by Horizon Europe, the Doctoral Candidate Santiago de León Martinez hosted at Kempelen Institute of Intelligent Technologies in Slovakia who explores the potential of eye tracking not only in theoretical research but also in applied, real-world contexts. He collaborates with industry partners to see how eye tracking can inform product design, user behavior analysis, and technological innovation. He reflects on the contrasts and synergies between academic and industrial research environments, the value of interdisciplinary collaboration, and how these experiences are influencing his research trajectory.
What kind of industry collaborations have you had so far, and what were your tasks?
I had the opportunity to work with 3 industry partners within the Eyes4ICU doctoral network: Eye Square, UI42, and Blickshift Analytics. Eye Square is a German UX research and market research company using eye tracking (and other methods) to understand how users/consumers interact and respond to platforms or services. Eye Square along with researchers at the University of Amsterdam’s IRLab led by Maarten de Rijke, collaborated with us to perform a large eye tracked user study of 87 participants to see how users navigate carousel (Netflix-like) interfaces. In addition, Eyes4ICU partner Blickshift Analytics also participated and helped with the processing and analysis of the gathered eye tracking data with their excellent software for eye tracking analysis. Finally, my collaboration with UI42 involved using my research expertise and even the results from our user study to help them optimize their recommender systems.
How did the work environment and research approaches in industry differ from your academic setting?
The collaborative work on the user study is one of the essential pieces of the PhD project and as it was led by myself at KInIT, so for me that was just standard academic research along with administrative work to connect everyone and make sure the study is running correctly. Fortunately, I had the opportunity to apply what I had learned with UI42, especially the insights from the user study of how to better design carousels. I would say the main difference between the industry setting with UI42 and academic research is the line when you say what is good enough or how you measure the costs/benefits of implementing or testing certain methods or techniques. In academic research, we need to be as thorough as possible and create environments where you have controlled tests so that you can rely on your results. In industry, especially with recommender systems, it is very difficult to have a controlled environment as you are running a live service that is impacting users. Thus any changes need to be effective and improve your system in a measurable and significant way. We are just as thorough with my work in UI42, but at the same time there are instances where you decide to pivot to work on another area that may be more effective for the system. This is in contrast with my research where I will follow a thread until its end to really figure out and answer the research question that I have.
Were there any challenges or limitations you faced when applying academic methods or tools in a real-world setting like at UI42?
Evaluating the changes you make to a live system that lots of users interact with is a big challenge. We want to improve the recommender system for the user, but how do we measure that improvement? The reality is there are so many different ways to measure improvement and you can’t optimize all of them at the same time. Also our changes affect a live service and changes aren’t guaranteed to improve the system, and at the same time a drop in performance can be due to other external factors. Effectively it’s a big balancing act of trying something new out and seeing how long we want to test it, while trying to have the best recommendations at any one moment. In practice, this means that we do a lot of A/B testing (trying solution A for a period of time and comparing it to solution B that was run for a similar period of time). I like to think of it as research on the fly. Industry work uses the same skills and techniques that I have developed as a researcher, but there’s the additional challenge of a less controlled environment and time or cost trade off that is much more pronounced.
In your opinion, what makes a research collaboration with industry successful?
Research and industry often have aligned goals. In research, the goal is to answer some questions and discover knowledge that usually will improve some system. Industry is more directly about improving the system. So the most important thing is to be on the same page about what we are improving and how it is improving. For that reason, I would say that communication, transparency, and trust are key for research and industry collaboration. For example, I communicate how confident I am of a certain change in the recommender system and when I am not sure I am open about that. This allows us to build a relationship of trust and at the same time serves the purpose of helping prioritize changes and see how long we are willing to test them out.
What advice would you give to other DCs heading for industry collaborations?
Communication with your industry partner is key. Communicate and find the instances where your research goals align with the industry goals because they often do. Then use your research skills in the direction of the aligned goal, but remember that you may need to be more flexible or adapt your direction at certain times, especially to suit the goals of industry.
What are your next steps after the PhD and the current projects?
I have really enjoyed my experience working with industry and would like to explore it more. I have worked as an academic researcher since I was 14 years old, so exploring this new (but familiar) direction of research in industry sounds very exciting for me. Regardless of whether I come back to academia or stay in industry, it will be a great learning experience that will help me grow as a researcher.
If you are interested in the open access publications that resulted from the collaboration see: