API gallery page for Paper List Widgets

[wammar]

Next steps:

• [ ] WIP: Iterate on design.
• [ ] Decide if we're going to include photos for this quarter based on the private APIs.
• [ ] Design a thumbnail for the top-level gallery page.
• [ ] Launch.

[nikett]

Hi Waleed, paperlist is pretty much done (thanks to Matt for helping with the looks, and me and intern already added caching as well to limit number of api requests). I have just been struggling with the availability of the image server that Luca set up. The server does the following: given a list of paper ids, it returns the image url. Is there a way to make that server a little more reliable (currently it fails on about 99% of the requests). I asked Luca who told me that it can be fixed once Luca is back.

[nikett]

I added a branch with paperlist. This is work that I started and then together with my intern Yash Kumar Lal for javascript and with Matt L for a nice UI. @MLatzke you can look at the branch https://github.com/allenai/s2-folks/tree/paperlist

[MLatzke]

Thanks, @nikett, I'll take a look!

[nikett]

Do you know if the thumbnails links from a paper will be made available via s2 api?

[cfiorelli]

@MLatzke / @nikett , I understand this ticket represents work to publish "getting started" code for different use cases? I only briefly poked around the paper list repo - looks cool. Any news on status?

[nikett]

I am waiting on this TODO: "Decide if we're going to include photos for this quarter based on the private APIs."

Everything else is good (and this is in use on my own homepage)

[cfiorelli]

@nikett The link to paperlist seems to just go to our s2folks repo. I was hoping if work on this is done we can add it to upcoming December API newsletter and gallery page, if that seems appropriate to everyone here?

[nikett]

I can finalize this and would love to see it in the gallery page. l have been waiting to hear back from S2 about whether the image thumbnails will be present in the S2 api? Because that simplifies my code and makes it robust.

On Sat, Dec 9, 2023 at 10:27 AM Christopher Fiorelli < @.***> wrote:

@nikett https://github.com/nikett The link to paperlist https://github.com/allenai/s2-folks/tree/paperlist seems to just go to our s2folks repo. I was hoping if work on this is done we can add it to upcoming December API newsletter and gallery page, if that seems appropriate to everyone here?

— Reply to this email directly, view it on GitHub https://github.com/allenai/s2-folks/issues/72#issuecomment-1848607520, or unsubscribe https://github.com/notifications/unsubscribe-auth/AFLLHUPATZRAI76PFXA6OWLYISUSTAVCNFSM6AAAAAAXUSRAMSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTQNBYGYYDONJSGA . You are receiving this because you were mentioned.Message ID: @.***>

-- Thanks, Niket

[cfiorelli]

Image thumbnails in the API -as in will we add a field to the api to allow for url to an image thumbnail, or? Sorry I do suspect I am not understanding. If you like we can hop on a call

[cfiorelli]

@rodneykinney - Can you give a sense of how much effort it would be to provide a field in the public API which returns links to paper thumbnails? This feature would Niket with the final piece he needs to build a code example which users would place on their website to show a feed of papers they're on.

cc @Jgordo72 - I found the below comment from you regarding licensing. Can you confirm that if a Springer abstract were somehow visible in the thumbnail via public API we would be OK?

If the image is available via the S2 page, then I have no issue with the redistribution via the API. We cannot pull images from some PDFs due to licensing restrictions from our publishing partners.

[rodneykinney]

By "thumbnail" do mean URLs for images of the paper's figures? That's relatively easy to implement. I'd call it a Medium.

[cfiorelli]

@rodneykinney - I understand we need to get a sense of priorities as we move into start of 24' but for the moment here's some clarifying notes from Niket. (same notes from our call today.)

https://www.semanticscholar.org/paper/RL4F%3A-Generating-Natural-Language-Feedback-with-for-Akyurek-Aky%C3%BCrek/ebf35cef5c249d90b40043fffa41f8802c27f132 For instance in this paper, figures and tables that are displayed. And I pick the first thumbnail

I just want to show the first for now because typically the first image is a representative I just need urls and not the images as such And I don’t need the table Essentially first image is all I need for now. Maybe in the long term it will be a good machine learning problem to decide a representative image given a list of images and tables from a paper. If it is easier to simply return all the tables and figures urls from the paper then that’s also fine as for the resolution - it could be a thumbnail or a figure. Thumbnail urls are actually even better for my use case

[Jgordo72]

@cfiorelli - There should not be an abstract visible in any image within the PDF. If it is visible in a thumbnail, then we would need to remove it to adhere to the Springer contract terms.

[cfiorelli]

Moving this to 2024 prioritization review !

thanks everyone