6 ms·
You should look at DocumentCloud and see if it would be helpful for hosting and managing the documents you're finding. http://www.documentcloud.org/home http:/
by donohoe 13y ago
You should look at DocumentCloud and see if it would be helpful for hosting and managing the documents you're finding.
http://www.documentcloud.org/home http://www.documentcloud.org/home
https://github.com/documentcloud https://github.com/documentcloud
It might be ideal for your needs. The project uses Tesseract‘s OCR to read documents and OpenCalais to pull data from them, including key people, places, and dates.
- jdimov 13y agoI doubt Tesseract will do well with handwritten Ukrainian... and I am positive OpenCalais won't detect anything in that language.
- pudo 13y agoThe project initially used DocumentCloud, but I guess that's built for scenarios which are a little less horrible than single-page JPEGs coming in via FTP. Tesseract has been working beautifully, we've got the first 1500 documents sent through it. Unfortunately, we don't know the language of each document (trying to get up some crowdsourcing for that), so each gets sent through three times (eng, rus, ukr). Finally, all versions are indexed in ElasticSearch. If anyone has a neater way of doing this (e.g. a good, Python-based language detector), please shout!