Open allanlasser opened 8 years ago
So when someone appeals, it should be possible to capture this if they input which exemption was invoked that they are appealing. If there is no appeal, we could still have something on the site for users to manually submit which exemptions were cited for the request, although this seems like it would have lower compliance.
We could create a task for every rejected or partial request, and have staff enter the exemptions, but that would create too much work for staff.
We could create those same tasks and MTurk it.
We could try automating it, but I think this would depend on how good the
OCR is, as I expect all the relevant text to be in a PDF and we would
pretty much just be looking for exact matches ie (b)(6)
, so if OCR is
good this could work really well.
Some combination of the above.
On Thu, Sep 8, 2016 at 4:42 PM, Allan Lasser notifications@github.com wrote:
Assigned #1055 https://github.com/MuckRock/muckrock/issues/1055 to @mitchelljkotler https://github.com/mitchelljkotler.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/MuckRock/muckrock/issues/1055#event-783044426, or mute the thread https://github.com/notifications/unsubscribe-auth/AAIToopt5TK3AjOwzoTDG4Gpcna8tYL2ks5qoHMpgaJpZM4J4bkt .
Collecting a few example requests and what exemptions were invoked might give us a better idea at which automation techniques might be most viable.
On Fri, Sep 9, 2016 at 1:20 PM, Mitchell Kotler mitch@mitchellkotler.net wrote:
So when someone appeals, it should be possible to capture this if they input which exemption was invoked that they are appealing. If there is no appeal, we could still have something on the site for users to manually submit which exemptions were cited for the request, although this seems like it would have lower compliance.
We could create a task for every rejected or partial request, and have staff enter the exemptions, but that would create too much work for staff.
We could create those same tasks and MTurk it.
We could try automating it, but I think this would depend on how good the OCR is, as I expect all the relevant text to be in a PDF and we would pretty much just be looking for exact matches ie
(b)(6)
, so if OCR is good this could work really well.Some combination of the above.
On Thu, Sep 8, 2016 at 4:42 PM, Allan Lasser notifications@github.com wrote:
Assigned #1055 https://github.com/MuckRock/muckrock/issues/1055 to @mitchelljkotler https://github.com/mitchelljkotler.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/MuckRock/muckrock/issues/1055#event-783044426, or mute the thread https://github.com/notifications/unsubscribe-auth/AAIToopt5TK3AjOwzoTDG4Gpcna8tYL2ks5qoHMpgaJpZM4J4bkt .
Is this something we are still interested in?
We have
InvokedExemptions
to track how and where agencies use exemptions, but we currently have no way of capturing or tracking this information besides having a staff member manually create each instance. I don't believe that actually recording exemption use was something that was documented or spec'd when we started this project, so it's still an open question.