Transition BioPortal from GA4 to Cloudflare Analytics

[jvendetti]

We're replacing Google Analytics (GA4) with Cloudflare Web Analytics as the source of page view data for BioPortal.

This issue serves as a high-level tracker for work taking place across multiple BioPortal libraries. Individual changes may be implemented in their respective repositories and linked back here.

Tasks

• Cloudflare analytics data collection implemented in ncbo_cron (PR #99)
• Update the REST API to source Cloudflare data
  • ncbo/ontologies_linked_data@b28f350
  • ncbo/ontologies_api@ed2ea2d
• Add email notification on data collection failure (see PRs #107 and #250)
• Augment domain filtering to include BioPortal slices
  • ncbo/ncbo_cron@ad461a6
  • https://github.com/ncbo/bioportal_config/commit/3dce9bc18f4ce7c83cb5e4d98c19b81cd1113e37
• Generate ontology rankings independently of GA4 analytics collection job

Notes

• Data is collected nightly and stored in both a JSON file and Redis
• Page view counts are derived from request-level metrics using Cloudflare's GraphQL API
• Bot filtering is limited based on the plan
• The GA4 analytics code will remain in the codebase to support OntoPortal virtual appliance deployments

[jonquet]

Hello. Good news. But is Cloudflare Web Analytics GDPR compliant?
I mean, many of our use of OntoPortal are in EU and GA was marked there as non RGDP compliant.
If we change, we should take this into consideraiton.

@asbage any opinion?

[asbage]

@jonquet yes, i am agree, it's not RGPD compliant, there is Matomo https://fr.matomo.org/blog/2022/03/gdpr-compliant-web-analytics/

[jvendetti]

We chose Cloudflare Web Analytics partly because it’s privacy-friendly, but also because we’re already using (and paying for) Cloudflare to protect BioPortal from bot storms and heavy scraping that were seriously impacting the site’s performance. Their bot mitigation has been essential for keeping the site stable, so using their analytics tool was a natural fit—it let us avoid adding another vendor or extra setup.

[jonquet]

I see the point of course. Good to know the solution you use against bots s we experience the same issues.

However, I am concerned for the "full open source" argument of the OntoPortal software. I personally can cope with the idea that some part are not free. But (when we change the triple store) multiple members have expressed concerns related to the fact that it would be good to keep the software fully free for all the small use cases and structures who like to use it without a big infrastructure and or organisation to pay in the back.
What do you think ?

[alexskr]

Hey @jonquet ,

Totally get where you’re coming from — keeping OntoPortal fully open and usable for small setups is important, and we’re not looking to change that.

The Cloudflare analytics integration is completely optional. It's not a replacement for Google Analytics, and nothing depends on it being enabled. We added it mostly because we were getting hit hard by bots, and Cloudflare's mitigation has actually been reasonably effective for us, but I understand that Cloudflare isn’t the right fit for everyone. Other teams may have different requirements, constraints, or preferences based on their specific needs or policies.

Their Web Analytics works on the free tier, so there’s no cost to use it in smaller deployments. That said, yeah — Cloudflare has its issues, especially with locking certain features behind paywalls. We’re not thrilled about that either.

This addition just extends the existing setup — similar in spirit to the old GA integration, but a bit cleaner and covered with unit tests. It doesn’t increase complexity unless someone opts into using it.

Matomo is also a great option. If we ever move away from Cloudflare or want to offer a more self-hosted alternative, we’d definitely be open to supporting that too.

So overall — no change to the philosophy. Just giving folks another tool if they want it, but keeping the core stack fully open and free.