Radboud Open Science Inspirator: Anas Maya

Practicing Open Science requires putting a generous amount of time and effort into producing a reusable dataset. To me, it all starts with one key question: Could my data also be valuable to others? If the answer is yes, then I better not keep it to myself – I should make it usable for others. This can be a challenge because I have to imagine how other scientists, whom I do not know, could reuse my data.

Successful reuse of data requires that it is described and structured in a way that makes it understandable for others and potentially combinable with similar datasets. Imagine research teams in different countries working on the same type of diabetes, each using their own terminology and languages. You need to be multilingual and aware of decisions of all teams to be able to map concepts onto each other. That’s why datasets need to be populated using standardised terminology. Without that standardisation, data is rarely truly Findable, Accessible, Interoperable and Reusable (FAIR).

'For me, it all starts with one important question: Could my data also be valuable to others?'

Alongside my role as a data steward at the Faculty of Arts, I’m also part of the Radboud Healthy Data Programme – a joined program between the university and the Radboud university medical centre. The aim is to develop a digital infrastructure that connects research (health) data with AI expertise on the Radboud Campus. To get a sense of how data sharing is going across campus, we ran a survey among researchers at Radboud University and the medical centre. A total of 518 researchers completed it. One of the findings: many researchers struggle with the idea of controlled vocabulary. In fact, over half of respondents weren’t quite sure what it even means. In the natural and life sciences, such standardisation is already fairly well established. In the social sciences and humanities, awareness is growing, but documentation is still often ad hoc and geared mainly towards human interpretation, as opposed to machines as well. So there’s still a lot of ground to cover.

Another key aspect of making data truly shareable is documentation. Many researchers believe during their project that they’ll remember everything or that all procedures or data elements are self-explanatory, which is far from the truth. Think of those vaguely named files on your computer – a few months later, even you don’t know what they are anymore. The most important step? Start thinking about this early on. 

Yes, documenting takes time, and using standardised vocabulary is not always straightforward. But the payoff is significant: well-structured data is easier to understand and reuse – also for AI applications – enables collaboration, speeds up research, and greatly increases the overall impact of your work.

Anas’ Open Science Tip

Don’t see your dataset as a single, indivisible entity. It’s not all or nothing – you don’t have to publish all your data exactly as it was collected. For example, instead of sharing data from 100 participants, you might share data from just 50. Or instead of raw audio recordings, you might provide only the transcripts. You can also anonymise, pseudonymise, or encrypt data. You can be flexible in how you choose to share.