srinivvenkat
commented
10 years ago I think there is no good, dependable OCR software that can do this task. And a dedicated team to just do this, is not feasible.
I just thought crowdsourcing it is the best way, if we are expecting more such pdf documents. In the long run, one could also decouple the data-collection, curation (tasks which can be crowdsourced to non-math, non-coders), from the modeling, analysis work. To help the crowd, can migrate to a less-geeky google docs kind of alternative.
Just giving a try. I collated the table pages alone in Guinea dataset pdf, and have created a google spreadsheet with the pivot column/row information (from 26th Aug onwards, the format is the same). It is accessible at ( bit.ly/ebola_guinea ). Have also added the 16th Sept, and 1st Oct .csv information. A few moderators could proof-read and 'freeze' cells which are confirmed (revision histories help too).
We've seen it on Wiki. We've seen it on reddit. Can we expect the Internet to do its magic here again?
tc-mccarthy
cmrivers
olarosling
Figure out a way to make clear which Guinea situation reports have already been converted to PDF, and which still need to be done. I want to keep the PDFs in the repo even after they are digitized, since they are not available easily available online like the SL and Liberia sitreps.