R script for a human-assisted process to screen GDELT for atrocities

gdelt.atrocities.daily.screen

# Clear the workspace
rm(list=ls(all=TRUE))

# Set working directory to location of extracted daily update and .txt file with column headers
# setwd("PUT YOUR DIRECTORY NAME HERE")

# Load function to return subset involving violence against civilians
extractor <- function(date) {
     filename <- paste(date,"export.csv", sep=".")
     dailies <- read.delim(filename, header=FALSE)
     header <- read.delim("csv.header.dailyupdates.txt", header=TRUE)
     names(dailies) <- names(header) 
     z <- subset(dailies, 
                      ( EventCode==180 | EventCode==1823 |
                        EventCode==190 | EventCode==193 | EventCode==194 | EventCode==195 |
                        EventCode==200 | EventCode==201 | EventCode==202 | EventCode==203 |
                          EventCode==204 | EventCode==205 ) &
                      ((Actor2Type1Code=="CVL" | Actor2Type1Code=="OPP" |
                          Actor2Type1Code=="EDU" | Actor2Type1Code=="LAB" | 
                          Actor2Type1Code=="REL" | Actor2Type1Code=="HLH" |
                          Actor2Type1Code=="REF" | Actor2Type1Code=="MED" ) |
                       (Actor2Type2Code=="CVL" | Actor2Type2Code=="OPP" |
                          Actor2Type2Code=="EDU" | Actor2Type2Code=="LAB" | 
                          Actor2Type2Code=="REL" | Actor2Type2Code=="HLH" |
                          Actor2Type2Code=="REF" | Actor2Type2Code=="MED" )) &
                      ((Actor1Type1Code=="GOV" | Actor1Type1Code=="MIL" |
                          Actor1Type1Code=="COP" | Actor1Type1Code=="SPY" |
                          Actor1Type1Code=="REB" | Actor1Type1Code=="SEP" ) |
                       (Actor1Type2Code=="GOV" | Actor1Type2Code=="MIL" |
                          Actor1Type2Code=="COP" | Actor1Type2Code=="SPY" |
                          Actor1Type2Code=="REB" | Actor1Type2Code=="SEP" )) ) 
     z1 <- subset(z, select=c(GLOBALEVENTID, Actor1Name, Actor2Name, EventCode,
                              ActionGeo_FullName, SOURCEURL))
     z1$ushmm.screened <- 1  # Marker for events we reviewed
     z1$ushmm.eoi <- 0  # Marker for events that look like atrocities
     return(z1)
}

# Extract the subset of interest
today <- "yyyymmdd"  ### INSERT DATE OF RELEVANT DAILY UPDATE ###
set <- extractor(today)

# Open subset in table and manually code event marker (ushmm.eoi)
# based on criteria specified in accompanying manual
fix(set)

# Write out the results as a .csv
write.csv(set, sprintf("%s.atrocities.csv", today), row.names=F)