7th Open Science Practitioners Forum

Name: 7th Open Science Practitioners Forum
Start: 2026-04-15T16:00:00+02:00
End: 2026-04-15T18:00:00+02:00
Location: No location set

Wednesday 15 Apr 2026, 16:00 → 18:00 Europe/Zurich

Surveys

Open Science Community Engagement Feedback

67317341543

Merten Dahlkemper

Hide

Minutes

If you haven't, subscribe to OSPF e-group: https://groups-portal.web.cern.ch/group/os-pf/subscribe
Snakemake hackathon took place: https://indico.cern.ch/event/1574891/timetable/?view=standard
Future meetings (tentatively):
- Trainings in June
- Open Science Infrastructure in September
- Open Source HW&SW at particle physics labs in November (in-person event, collaboration with OSPO)

Zach about Open Infrastructure meetinfgs: Hepdata, Swan, binder, collab etc (cloud infrastructures that support Open Data), EOSC infrastructure. How broad should this be?
- Could be split: Inward looking vs outward looking, or documents (INSPIRE, cds, indico) vs. data (hepdata etc)
Giacomo endorses the idea to have a joint event with OSPO

New Open Science website is currently being developed and to be published soon: Beta-testers needed
Community Ambassador network being launched: Get in touch if you want to share your excitement for Open Science
In-person OSPF should become a regular event: Feedback needed as to who would like to join
Indico survey for feedback

ATLAS can share a lot of stories, but it should be as clear as possible what the ask is
For training it needs to be known who the target audience is

Full chain of tools for Open Data from no-code solutions to OD for research
Collecting projects on how OD is used
Regional usage patterns monitored via monit-grafana: https://monit-grafana.cern.ch/d/da06d76c-24f0-4d23-b51e-da08d36c4ece/welcome?orgId=93
In 2026 US use went down, and Brazil went up, also Africa got some uptake
OD Tutorial (https://indico.cern.ch/event/1564767/): tried to cover all audiences, won't be repeated, rather online tutorials for specific audiences in 2026
Big takeaway: We need to understand the audience
Upcoming white paper on event generation data
proof-of-concept: Agentic workflow to easily generate output from the Open Data from natural language

How do you collect the materials people use
- Reaching out to individuals, word of mouth, googling
- Can be tricky to track, as often data is only downloaded once and then used regularly in education
What about the Masterclasses?
- Masterclasses are very well known and widely used, so it's hard to replace.
- Needs more discussions with IPPOG, maybe will be partially replaced
Why using atlasopenmagic and not something from the open data portal?
- Long term vision would be to have something universal for everyone

It was clear the data needs to be preserved, but who would use open data?
First use case by theorists: Had significant challenges with the file format
Eventually had the first OD workshop for theorists in 2020
So far in total 6 workshops: https://cms-opendata-guide.web.cern.ch/cmsOpenData/workshops/
Ratio participants to registrants is relatively low
Workshops so far have received good feedback (>90% recommendation rate)
Data format has now changed to NANOAOD from AOD in the beginning, which does not require CMS specific software
Next workshop: 28-30 July 2026 at University of Notre Dame, USA, focus on educational use-cases targeted to teachers

In ATLAS also issues with no-shows, maybe worth considering a small participation fee so people are committed
What about the drop-out rate over the week as material gets harder
- Decline was not super high
- Had large registration numbers in the beginning, but then many didn't show up. However, actual participation seemed to be quite constant
- Some wanted to get a certificate for university. For this it should be necessary to go through exercises
Which datasets do you use in the CMS workshops? And do participants actually understand an actual analysis?
- Hopefully the students should understand the whole analysis, at least that's what the material is presenting
- Support is there via the OD Forum

Ntupling Service is now released since February, no more need to know LHCb tupling software
Ntuples are created on Grid and can be downloaded by user
Access to 4 PB of Run 1 & 2 data
Needs some guidance as first-time user: https://lhcb-opendata-guide.web.cern.ch/ntupling-service/
Currently working on implementing more examples, to be published by May
Only 2 people actively working on Open Data in LHCb
Several presentations at conferences etc.
Answering questions from users
So far 14 requests, various use cases (CP asymmetry searches, fitting algorthms, educational/learning root)
So far 4 TB od OD has been produced
Will organize future LHCb OD events

What's the background of the people placing the requests?
- Mostly either theorists or high school students
Are there restrictions for LHCb members to use the OD? It might be interesting to create educational datasets
- For educational datasets, this is used also by LHCb members; for physics publications it's not allowed to use OD for LHCb members
How is this advertised?
- On the outreach website and conferences; not via Social Media
- Could be worth putting in physics talks

Data format has been changed after run 2 to AO2D, so the run 1 and run 2 data has to be converted into this format to be shared as OD
currently old format data (7.6 TB) still in OD Portal, but probably cannot be used anymore
2015 Pb-Pb data (62 TB) has been released as educational data in March
In the next years much more data is being released to hit the release targets, expect to upload the converted run 1 data by the end of the year
O2OpenAccess is the software repository for ALICE Software: https://github.com/AliceO2Group/O2OpenAccess
Software is the same software for internal use and open data

There are minutes attached to this event. Show them.