Laying the Foundation for FAIR Scientific Publications

Wednesday - May 10 2023, 22:11 UTC - 1 year ago

Knowledge Pixel is a software company that developed a framework for publishing scientific findings in an actionable and human-readable manner. The company has used FAIR principles to make sure the research data is organised in a machine-interactive way, with Nanopublications at the core of the mission. The duo have teamed up with academic publishers Pensoft and IOS Press in order to implement their vision.

Standing at the dawn of Artificial Intelligence (AI)-powered reality, we can hardly think of anything the exponentially advancing technologies would struggle to digest. A critical exception would most certainly be scientific knowledge, which has long been accumulating across disciplines, fields and academic publications in the form of extensive narratives based on data and observations.

Until AI technology "learns" how to interpret complex scientific literature and evaluates the data, methodology and evidence behind it, it is our responsibility to make sure what we know today is optimally available to the computer algorithms. In turn, those algorithms would be extremely helpful in assisting researchers to build on the knowledge of yesterday by delivering the right information at the right time in a ready-to-use format.

This is why the team behind Knowledge Pixels, a recent startup that develops software and services, devised a framework to publish scientific findings in a way that is simultaneously human-readable and machine-actionable. To do this, the duo teamed up with forward-looking scholarly publishers Pensoft and IOS Press to implement its goal.

The collaborators share a common vision and mission founded on the FAIR Principles, that is ensuring scientific outputs are Findable, Accessible, Interoperable and Reusable. To put it simply, you can find a FAIR publication on the Internet in a few clicks, and so can a computer algorithm mining the Web. Then, both you and the algorithm will be able to integrate that FAIR item into your own output almost instantaneously.

The thing is, the academic community has been talking about sharing scientific findings in a FAIR manner for quite some time, but such a publication workflow is yet to see the light of day.

"The way how science is performed has dramatically changed with digitalization, the Internet, and the vast increase in data, but the results are still shared in basically the same form and language as 300 years ago: in narrative text, like a story. These narratives are not precise and not directly interpretable by machines, thereby not FAIR. Even the latest impressive AI tools like ChatGPT can only guess (and sometimes 'hallucinate') what the authors meant exactly and how the results compare," say Philipp von Essen and Tobias Kuhn, the two founders of Knowledge Pixels.

When asked why there are no FAIR publications of scientific results yet, Philipp von Essen states, "We use nanopublications as our core technology. The conceptual basis for nanopublications was laid more than 10 years ago, at the time when Semantic Web was a bit of a hype. It turned out that the technical implementation was not so easy and the concepts needed further development." .

"Only a few researchers pursued the ideas and continued to work on prototypes. One of them was Tobias Kuhn, and now we are ready to put nanopublications into scientific practice." .

When asked why FAIR publications are so difficult, Tobias Kuhn explains, "We want research findings to be FAIR, because they can then be automatically interpreted and compared by computers, thus multiplying the value we can draw from them. Yet, to make such findings fully interoperable, as the FAIR principles require, they need to be specified clearly and consistently." .

"For example, when we attributed certain temperature values to certain geographical locations, we needed to make sure all temperatures are in the same measurment scale, otherwise no computer algorithm would be able to interpret them." .