design top images mosaic, tree map or other display for topic top stories

[cindyloo]

No description provided.

[cindyloo]

in our tech call, I presented that we could run tensorflow predictions within javascript after we train a model via python or what-have-you. It seems we need to store/download the image and url for data preservation. This makes sense, however, however I believe we can still do lightweight, dynamic predictions within the front-end (utilizing a tensorflow/keras model trained in the back-end at specified intervals) IF that makes sense.

these predictions could be along several lines;
-pixel-wise similarity of image
-similarity of image detection (eg, flags, state houses, etc)
-similarity of facial detection (enables facial centering and cropping)
-similarity of facial identification (enabled id of who is in what pic)

On the other hand, perhaps we want to estimate the predictions once, and store that information as well as the url and image per top story.

[rahulbot]

I think we need to go back to basics to make these technical decisions. I've tried to write up a few of the key use-cases on the back-end project. Please chime in there so we can refocus on which research idea each tech solution might support more readily deploying: mediacloud/backend#708

[hroberts]

I agree that we need more work to figure out exactly what it is we are building.

…

On Wed, May 20, 2020 at 1:03 PM rahulbot ***@***.***> wrote: I think we need to go back to basics to make these technical decisions. I've tried to write up a few of the key use-cases on the back-end project. Please chime in there so we can refocus on which research idea each tech solution might support more readily deploying: mediacloud/backend#708 <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_berkmancenter_mediacloud_issues_708&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=0c5FW2CrwCh84ocLICzUHjcwKK-QMUDy4RRw_n18mMo&m=NF9rFg1e5Tp17ajYEVlA--A2bXxhf-PTWIxUFwFEkSo&s=v6TpQpcJkQsokdRAmuRZB2lps1JP46AxMOf2iXB-dnA&e=> — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_mitmedialab_MediaCloud-2DWeb-2DTools_issues_1814-23issuecomment-2D631634482&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=0c5FW2CrwCh84ocLICzUHjcwKK-QMUDy4RRw_n18mMo&m=NF9rFg1e5Tp17ajYEVlA--A2bXxhf-PTWIxUFwFEkSo&s=JiVkFy7t4aBMa0r0x7VKeTcbtQxI8W5mcflnI56KhIc&e=>, or unsubscribe <https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_notifications_unsubscribe-2Dauth_AAN66T4BWZWCHMFURF7VQY3RSQLNFANCNFSM4MYZYSXQ&d=DwMCaQ&c=WO-RGvefibhHBZq3fL85hQ&r=0c5FW2CrwCh84ocLICzUHjcwKK-QMUDy4RRw_n18mMo&m=NF9rFg1e5Tp17ajYEVlA--A2bXxhf-PTWIxUFwFEkSo&s=xHd0vflJVYB8qYUIJH45_xZeZZxXjsr7PZYDbTH5nxA&e=> .