Stream: community

Topic: Updating data on a recurring pipeline


view this post on Zulip Brian C. Keegan (Jun 20 2026 at 03:41):

Hi folks, newcomer to the community (and actually wrapping up at Harvard as visiting scientist!).

I'll be returning to Boulder in August and rebooting my lab group. I am interested in exploring the intersection of agentic AI, public interest data liberation, and digital archives.

One tool I'm prototyping built a CI pipeline for a daily scrape of my local police department's "Flock Transparency Portal." I'd like to archive and DOI-ify it while updating it at a regular interval. https://github.com/brianckeegan/boulderpolitics/tree/main/Part%2006%20-%20Flock%20ALPR%20contract

What advice, standards, docs, or examples would you recommend about archiving/DOI-ing a dataset that updates on a monthly/quarterly/annual frequency?

view this post on Zulip Brian C. Keegan (Jun 20 2026 at 03:43):

Feedback on these prototype skills also welcome:

https://github.com/CUPIDS-Lab/data-project-skill

https://github.com/CUPIDS-Lab/data-liberation-skill

view this post on Zulip Philip Durbin ๐Ÿš€ (Jun 20 2026 at 15:00):

Hi! Welcome! For now, I'll just say that I'm reminded of this issue, which you might find interesting:

Create a configurable, scheduled process for the BARI researchers to process and deposit ("stream") Analyze Boston data into their dataverse automatically. #4530

view this post on Zulip Johannes D (Jun 22 2026 at 07:22):

We also have some customers that update their data each night, which results on a lot of datasets versions. There is potential for quite some performance improvement but in general it does work as expected.

view this post on Zulip Philip Durbin ๐Ÿš€ (Jun 22 2026 at 13:37):

@Brian C. Keegan my colleague @Leo Andreev pointed to some concrete examples in a reply to "Institutional datasets with monthly updates โ€” looking for examples & best practice" at https://groups.google.com/g/dataverse-community/c/iSvAnq_H4sE/m/w6aOcjmnAQAJ

view this post on Zulip Philip Durbin ๐Ÿš€ (Jun 22 2026 at 13:44):

He points to https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/VH6GVH which now has 430 versions. This is quite a lot for Dataverse (it takes a while for the "Versions" tab to load) so you might want to be careful about the frequency of updates.

view this post on Zulip Oliver Bertuch (Jun 22 2026 at 14:24):

It might be interesting to take a look at what's going on under the hood there by extending the performance testing framework I contributed. I bet there are a few low hanging fruits to improve this.


Last updated: Aug 18 2026 at 08:27 UTC