Conclusions and recommendations
Our policy that requires researchers to disclose the reason why they may not be able to share their data openly, has brought to light that one-size-fits-all solutions are unlikely to be effective. The diversity of challenges and the varying timescales needed to address them call for a more nuanced set of approaches to foster open data sharing. Moreover, strict mandates alone won’t suffice, as our findings suggest that many challenges stem from a lack of resources or knowledge about how and where to share data, rather than a lack of willingness.
The insights gathered from our review of published data availability statements underscore the need for continued efforts to promote transparency and collaboration within the scientific community. While progress has been made, the relative infrequency of FAIR data sharing across the physical sciences remains a barrier that must be addressed through tailored strategies and supportive policies. By fostering a culture that values the FAIR principles, we can enhance the reproducibility and impact of scientific research. A key factor is making the process easier for researchers, ensuring they not only have the desire to share their data but also the knowledge and resources to do so effectively.
A key takeaway is that the diversity of data-sharing challenges means we must engage research communities from the outset and throughout the process in developing tailored solutions. Bringing researchers together with a broad range of stakeholders will ensure these challenges are addressed holistically. To achieve this, we also need to establish stronger communication channels between funders, institutions, publishers, and others involved in research data. As a society publisher it is our goal to represent the interests of our research communities in these discussions and to contribute towards an environment where data sharing is not only encouraged but seamlessly integrated into the research process. Recognising the different starting points and barriers research communities face is the first step toward fostering a culture of data sharing and unlocking the full potential of open science.
As a society publisher it is our goal to represent the interests of our research communities in these discussions and to contribute towards an environment where data sharing is not only encouraged but seamlessly integrated into the research process.
In support of these efforts, IOP Publishing recently launched an open access Machine Learning journal series offering new formats for data to be showcased, such as dataset, benchmark, and challenge articles. These new article types aim to incentivise and strengthen open data practices, promoting greater transparency and reproducibility in research. For instance, dataset articles provide detailed descriptions of research data, including how it was collected, processed, and relevant metadata. Challenge articles bring researchers together to solve specific problems by creating new algorithms, datasets, or workflows. Benchmark articles evaluate the performance of various models, algorithms, or software against a consistent problem or dataset. IOP Publishing remains dedicated to pursuing both independent and collaborative innovations that advance the broader adoption of open practices across the physical sciences.
“Open data has the potential to accelerate science, but only if we work together with research communities to overcome the obstacles that stand in its way. Through tailored strategies and collaborative efforts, we can pave the way for a future where data is shared openly by default, enabling scientific discovery and innovation to flourish.”
Daniel Keirs, Head of Journal Strategy and Performance at IOP Publishing.