Impact Bulgarian Demonstrator Dataset

From DigitWiki
Jump to: navigation, search

The Bulgarian ground truth produced by National Library of Bulgaria (NLB) in the frame of the EU funded Impact project consists of 1.276 pages in PAGE XML format with an accuracy of 99.95%, that is a maximum rate of error of 5 characters wrong on 10.000 pages.

The Impact Bulgarian GT is distributed under CC-By-NC-ND. It includes the following works:


Title Type Year Script No of pages Language

Biblioteka Sv. Kliment


Journal


1888


Cyrillic


214


Bulgarian


Bylgarska iliustracia


Journal


1880


Cyrillic


100


Bulgarian


Jenski glas


Newspaper


1889


Cyrillic


253


Bulgarian


Sborniche za spomen na 25-godishninata ot smyrtta na Levski


Book


1898


Cyrillic


58


Bulgarian


Spisanie Dennica


Journal


1890


Cyrillic


384


Bulgarian


Ugozapadna Bulgaria


Newspaper


1893


Cyrillic


39


Bulgarian


Zelokupna Bulgaria


Newspaper


1879


Cyrillic


104


Bulgarian