For the Ottoman archival documentation, we have recruited data scientists and developed our own expertise and capacity, to build a customised data infrastructure for the project. The challenge in this endeavour has been to design, create, constantly modify and update data entry templates in the front end for data inputters; and organise the dataset architecture at the back end; for data analysis. We opted for a relational database structure using Microsoft Access. The difficulty in using this and similar relational databases for data entry for historical sources is finding a method to customise data entry for multiple users without compromising data entry speed or accuracy. To reach this goal we, where possible partially automated, and semi-automated data entry. Having a permanent member in the team solely responsible for data entry, maintenance, and safe keeping enabled us to develop the digital research infrastructure we have been aiming for. With increased speed and reliance, we managed to expand the geographical territory for which we have been collecting demographic data from Ottoman population and economic data from Ottoman tax registers for the mid-nineteenth century considerably. Based upon the premise that Ottoman economic and demographic structure varied considerably in the 1840s, which build the base period for our long-term perspective, in order to be able to make inter- and intraregional comparisons, we have developed a regional comparative perspective and a concomitant data extraction principle and a three-tier regional method. First, we locate an urban centre as a primary location which is suitable to analyse long-term dynamics of industrialisation and urbanisation, which are the two main axes of our project. Secondly, we look for three towns (secondary locations) in the surroundings of that city and set boundaries for a region via including all villages (tertiary locations) in that region. In designing this three-tier regional perspective we qualify and disqualify possible candidates for primary and secondary locations via operating under multiple constraints: i.e. primary and secondary locations should be covered by the 1840s Ottoman population and tax registers. By focusing on primary and secondary locations we can analyse urban economic and demographic changes by bringing in selected urban census data.
After geolocating primary and secondary locations using GIS then we take the arduous task to find and geolocate hundreds of tertiary villages in our chosen regions which are listed in the 1840s Ottoman population registers with total number of households and males in ethno-religious sub-categories per village. After geolocating the tertiary locations, we finalise the boundaries for a region. In doing so we can create unprecedented and reliable population density maps of Ottoman regions preceding to population censuses. We use historical maps from census years to harvest shapefiles to be populated by locations and demographic data for villages prior to censuses. Furthermore, since our regions are delineated via GIS we can also calculate population densities beyond two-dimensional space and calculate surfaces of our regions using a Digital Elevation Model. These geospatial aspects allow us to calculate more realistic population densities by setting elevation limitations to settlements. We have digitised, spatially harmonised, and tagged data with historical maps corresponding to census years from population censuses of countries of Southeast Europe and Turkey using optical character recognition software and GIS. We have almost completed all censuses for Turkey for 1927-2000 and most of Bulgarian censuses for 1880s-1956 to acquire project specific demographic and occupational data at highest possible spatial resolution. Bringing pre-census and census data on regional scale allowed us to examine regional economic development in the long run.
