search

softcite / softcite_kb

A Knowledge Base for research software relying on large-scale text mining and curated knowledge sources
Apache License 2.0
15 stars 0 forks source link

Towards a demo screencast? #3

Open jameshowison opened 3 years ago

jameshowison commented 3 years ago

Would you like to collaborate on a script/storyboard?  Or would you just like to record a zoom session in which you demo that system?  Nothing more than you showed Karthik and I?  Then I could wrap that with some additional slides about how it was created?

  1. Start at listing of software packages, browse into a few.
  2. Search for a particular package, show browsing to papers that mention it.
  3. Demonstrate location of accompanying references for software packages
  4. Start at a particular paper, show browsing of software mentions.
  5. Show a few failures (there is work to do!)
  6. Perhaps show hitting the API via the browser (spitting out wikidata format data)
  7. Then hitting that same link via a jupyter notebook, parsing and graphing the data.

I would be excited to work on the last two, if that makes sense?  I guess that would require me either using it locally (docker?) or having the demo at a web accessible URL?

kermitt2 commented 3 years ago

Hi @jameshowison !

I send you the accessible online URL demo information by email.

Maybe it's a bit early to invest time on this kind of storyboard, the system is really WIP and for the frontend in particular is more for testing purposes, we can expect quite a lot of changes in the next week. So it means that a webinar/script might need to be updated soon.

So far I don't plan to have a Docker image for the project because the KB dump content is already close to 1GB and will increase significantly with the large scale mention extractions. So either we have a docker image without data, not really useful and usable without a lot of loading and ingestion, or we would have a mega-image. So I think the best for a jupyter notebook would be to use the online instance REST API.

jameshowison commented 3 years ago

Thanks! I hear you on the WIP, but nonetheless I think it would be great to have just a teaser for us to share, even if it's just exactly what you demo'd in the Zoom call?

We could even just do another call where we demo to Caifan and record that and overlay bits? Or just record yourself clicking around a bit and I'll do a voice-over? It's exciting stuff even if it isn't in a stable state at all.

I hear you on Docker image size, I will look for the online access and play around a bit :)

--J

On Fri, Apr 16, 2021 at 10:26 AM Patrice Lopez @.***> wrote:

Hi @jameshowison https://github.com/jameshowison !

I send you the accessible online URL demo information by email.

Maybe it's a bit early to invest time on this kind of storyboard, the system is really WIP and for the frontend in particular is more for testing purposes, we can expect quite a lot of changes in the next week. So it means that a webinar/script might need to be updated soon.

So far I don't plan to have a Docker image for the project because the KB dump content is already close to 1GB and will increase significantly with the large scale mention extractions. So either we have a docker image without data, not really useful and usable without a lot of loading and ingestion, or we would have a mega-image. So I think the best for a jupyter notebook would be to use the online instance REST API.

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/softcite/softcite_kb/issues/3#issuecomment-821255215, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAAWOUSG6WAPG4WGYABN2VDTJBJKPANCNFSM43ABI4XA .

kermitt2 commented 3 years ago

The zoom recording, with a bit of editing and/or voice over, should be a good and quick solution !