ux: better advertise LHCb Ntupling Service on the CERN Open Data portal (and vice versa)

[tiborsimko]

About

The CERN Open Data portal hosts close to 1 PB of LHCb open data from Run-1. However, for Run-2 and beyond, the official way of releasing LHCb open data is via the LHC Ntupling Service, allowing researchers to ask for custom LHCb ntuple production from over 10 PB of LHCb data.

The LHCb Ntupling Service is close to the official production release and it would be good to better interlink it with the CERN Open Data portal and to advertise it as an official way how to get LHCb open data from Run-2 and beyond.

This issue discussed various techniques how to achieve that, hoping we can select one or more (or all).

Technique 1: Improve LHCb content

The CERN Open Data portal contains "about" pages, and the page for LHCb is a bit underused: https://opendata.cern.ch/docs/about-lhcb

The page does not describe much what LHCb open content one can find where, and does not mention LHCb Ntupling Service either.

Proposal: It would be good to expand its content and to explain about LHCb Ntuplnig Service as the primary way of getting LHCb open data from Run-2 and beyond. Note that there is already cernopendata/opendata.cern.ch#3751 pull request waiting for review and merge.

Technique 2: Write announcement news

For important portal updates, we are writing news announcement text:

We can write a news article about the LHCb Ntupling Service with links to documentation and guide. This page will also appear in user searches etc.

Proposal: Let's reuse the same content that LHCb press release announcement might want to publish. This should be relatively easy and a no-brainer.

Technique 3: Put links in the headers

The current CERN Open Data portal header banner looks like:

The current LHCb Ntulping Service looks like:

A similar design of the page headers is not a coincidence, but the styles can be harmonised even better, and the two sites could be cross-linked.

Proposal: It would be good to make very visible cross-links in the headers, for example in a style like Google advertises Gmail and Images in the page header, giving on the top-right something like:

Technique 4: Search hints

If someone clicks on "Focus on ... LHCb" on the CERN Open Data portal home page, the person is brought to the search results page looking like:

Here the LHCb open data from Run-1 is presented, but there is no way of knowing that data from Run-2 can be obtained via the LHCb Ntupling Service.

Proposal: we could display a search hint to the user with a link to the Ntupling Service, for example:

If we do something similar for user searches when "LHCb" is used in the search pattern, or when "LHCb" is chosen in the experiment facet, then the user will be well guided as to how to get to LHCb open data.

We could make such a hint system generic and it could be useful for other types of queries, for example ATLAS education open data, etc.

Discussion

Any other ideas? @psaiz @jbenito3 @dillfitz @pietnogga @CameronMcClymont @michaelbuchar

[pietnogga]

Hi Tibor, thanks so much for preparing!

• I certainly agree about improving the LHCb display text. I think we should add some more information than written in feat(docs): update LHCb About and LHCb Getting Started pages opendata.cern.ch#3751
  (e.g. full physics program, important discoveries etc.)

• I really like Technique 4, which will be particularly nice the first months after release since it catches the eye very well. Maybe its worth thinking about including it as a permanent header in the list of the screenshot you showed! (I think this will be the primary way to obtain LHCb Run 1 open data as well)

• One general comment: we say we publish 10 Pb of LHCb open data (Run-1 and Run-2 combined). However, I am not entirely sure this is true, since we do not publish all data on the Ntupling service, but only data well suited for physics (for example, we do not include min bias data or VELO-open data, TurboStreams etc.). Let me check the total data volume of 2015-2017 data we release and adjust the text accordingly. Maybe @dillfitz knows more.

I will write here if I have any additional comments/ideas

[dillfitz]

Hi Tibor, this is excellent, thanks a lot! I agree with all of Piet's comments and can leave a few of my own.

• To follow up on bullet point 3 from Piet, I agree that we might want to adjust the 10 PB language, in particular since the 2018 data is not being released. Also even with the full data volume of run 2015-2017, it is hard to measure the effect of filtering out specific selections (the blocklist). Though I suppose, unless every line on a particular stream is blocked (which is extremely unlikely but I have not checked), then technically access to the full data volume of 2015-2017 is provided. @pietnogga, I would be curious to hear this number as well, it will probably be useful for our NSF grant.
• Really I am a fan of all of the techniques, but I realize that we will have to focus person power in particular areas. Do we need to pick and choose? Or can we make a list of priorities (e.g. I think technique 4 will yield the most immediate users and should be prioritized, but I really like technique 3, could it be addressed later?)
• Do you have a sense of engagement for your news announcements? Since we are already preparing a press release on the LHCb side, technique 2 seems like it would be pretty low effort.
• I am definitely in favor of implementing technique 1, as this will also help us get feat(docs): update LHCb About and LHCb Getting Started pages opendata.cern.ch#3751 merged -- I really want the updates to the usage guide for the Run 1 data already on the ODP to become more public as well. Of course we can take this opportunity to add information about the Ntupling Service.

[tiborsimko]

@pietnogga @dillfitz Thanks for the thoughts, the techniques are not mutually exclusive, I was actually thinking of implementing them all, they were listed in a sort of increasing implementation complexity, so to speak. WRT technique 4, I'll cross-check thoughts with @psaiz; perhaps the technique could be useful for other scenarios (such as gentle advertising of https://opendata.atlas.cern/ for ATLAS searches).

[dillfitz]

Hi @tiborsimko, anything we can do on this front? Is it feasible to implement techniques 3/4 in ~1 weeks time? Let me know what you think, and thank you!

[psaiz]

Hi,
Option 4 sounds a bit too farfetched. There is nothing in invenio that would provide this out of the box, and given that the other solutions seem much easier/straight forward, I would prefer to go with the other options.
Best,

pablo

[dillfitz]

@psaiz option 4 is by far the best advertisement of the service. I worry that a news release and an announcement of the service on the about page will be insufficient to advertise the LHCb Ntupling Service. This will be the main way that users interface with LHCb open data, so at the very least, it needs to be advertised visibly when users click "focus on LHCb" from the landing page. To me, option 4 seems like a practical way to achieve this. Otherwise, we need another way of advertising the service aside from just sticking it inside a single record. I should mention that I am also in favor of implementing option 3 if possible, though still feel option 4 would be the most instructive for users looking to analyze LHCb data.

[tiborsimko]

• Option 1 done and deployed on PROD. feat(docs): update LHCb About and LHCb Getting Started pages opendata.cern.ch#3751
• Option 2 done and deployed on PROD. feat(docs): add LHCb Ntupling Service release news opendata.cern.ch#3776
• Option 3 pending. For the Ntupling Service, we advertised the CERN Open Data portal in the web site banner rather prominently. For the CERN Open Data portal, this may be perhaps too much, so we may want to tune in down (see the original proposal), or find something more subtle on the home page).
• Option 4 pending. It would be really good to do this in near term. Starting from Run 2 data, the LHCb Ntupling Service is the official way how to get LHCb open data, so we should advertise it rather prominently when people are searching for LHCb data on the CERN Open Data portal... I agree with Dillon.