The project cleans, standardizes and combines historical databases on human rights violationsin El Salvador between 1975 and 1993.
The following databases were used:
- Rescate: Based on Tutela Legal denuncias, independent work
- CDHES-LE: contains Individual victim depositions
- United Nations: Comision de la Verdad (direct sources)
- United Nations: Comision de la Verdad (indirect sources)
- SJC:
This project is a collaboration of the University of Washington
Center for Human Rights (UWCHR) and the
Human Rights Data Analysis Group (HRDAG).
These are relative to SV-history/
individual/: contains the individual databasesmultiple/: contains databases that have been merged for analysis (naming convention db1+db2.ext); PB comment: this category level is unnecessary. To concat all the files together, usecompare/CDHES-le+Rescate+UN+UN-indirect/importmatch/: to deduplicate all the violationscompare/: to compare (usually for descriptive purposes) raw datasets to each other: how many detentions does CDHES-le contain vs how many in Rescate?analyse/: contains analyses and results for various modelswrite/: contains files used for writing reports, papers, etc. (usually in LaTeX)
{database_name}/: directory for a given databaseshare/: files that are commonly used across databases
These are a good typical list but vary by category. Each task is explained in a README file.
import/: convert database file format to csvmerge/: link tables together using matching unique keysclean/: fix systematic and ideosyncratic value errorscanonicalize/: recode values in a column to some standard set for consistency within and across databases, particularly for dates, locations, violations and perpetratorsexport/: write versions of database, renaming columns for consistency across databases and dropping unnessary columns, based on specific needs
input/: contains raw import files or symlinks of file outputs from a previous task directorysrc/: contains scripts and Makefileoutput/: contains output files generated fromsrc/hand/: contains files or symlinks with data created by hand or using notebooks, typically csv tables that pair raw values and corresponding canonical values for dates, locations, violations and perpetratorsnote/: contains Jupyter notebooks that test code, help generate files for hand or explain the logic of complex cleaning tasksfrozen/: contains input files that have been modified by hand to correct issues that are not fixed with code; where they exist, they should be used in place of files ininput/doc/: contains documentation