Data extraction with LLMs. Is a data publication still necessary?
Two weeks ago, FAIRmat published a short report about their latest FAIRmat User Meeting, which was all about using artificial intelligence (AI) and machine learning (ML) approaches for next-generation materials discovery. In that report, they promised to publish the recordings of the talks of the event. Well, they delivered, with 11 new videos on their YouTube-channel (you can find the link to the playlist below) as well as an overview on the news page of FAIRmat.
Link to the YouTube playlist on the FAIRmat YouTube-channel
Link to the news page of FAIRmat
In the News Digest today, I only want to highlight one recording, and it is the hands-on workshop by Mara Schilling-Wilhelmi with the title "Large Language Models for Scientific Data Extraction". This video is worth your time if you have ever asked yourself: "With the development of LLMs do we actually need to put so much effort in publishing our data alongside the research article (e.g. in a repository). Can't the LLMs just extract everything?". It is not only shown how to extract data with LLMs (and lets you do it yourself) but also shows how much effort it takes to get data out of a research paper and ensure it is correct even with the help of LLMs.
What actually is the PSDI?
Are you new to the News Digest and have never heard of PSDI? Or you know roughly what it is, but an update would be nice. Check out the latest upload to the Zenodo community of PSDI. A nice short presentation giving an overview and introduction to PSDI.
Community engagement in developing a sustainable data infrastructure for physical sciences [Presentation]
Link to presentation on Zenodo with DOI: 10.5281/zenodo.15849153
How to access services and tools of PUNCH4NFDI? Add this to your resource collection.
Slides from a talk from 2022 about the plans to the Science Data Portal (SDP) and Digital Research Product (DRP) were uploaded to Zenodo. If you have never heard of SDP and DRP of PUNCH4NFDI, take a look. But what I also want to share something else. I was browsing on the PUNCH4NFDI webpage and stumbled on the Results of PUNCH4NFDI webpage (see link below), where all results are published (so tools, services and more). You'll want to add this to your resource collection, it might come in handy in the future. Which services overviews do you like to use or forward to your researchers? Let me know in the comments.
Link to the results of PUNCH4NFDI webpage
Science Data Portal and Digital Research Product of PUNCH4NFDI [Presentation]
Link to presentation on Zenodo with DOI: 10.5281/zenodo.16637860
Outro
That is it for today, as always, I hope you found at least one useful thing in this News Digest. Thanks for reading and see you next week!