MAPS-OF-WOTR

I. Different corpora in /corpora/wotr:

1. 'cwred-20150415': CWRED corpus of Civil War military actions, from
   Scott Nesbit.

2. 'dsl-080414': Corpus of emancipation and similar events.

3. 'volspans-predicted-wotr-*-predicted-deg*', e.g.
   'volspans-predicted-wotr-may-18-715pm-predicted-deg1': Predicted volume
   spans with predicted coordinates from WOTR, using annotations from the
   given date and running the predictions using Naive Bayes with uniform
   grid size as given.

4. 'wotr-*-60-20-20', e.g. 'wotr-may-18-715pm-60-20-20': 60/20/20 split
   of annotated WOTR spans for the given date.

5. 'wotr-*-100-0-0', e.g. 'wotr-may-18-715pm-100-0-0': 100/0/0 split of
   annotated WOTR spans for the given date.

II. To generate a KML map of the distribution of a corpus, e.g.
'may-18-715pm-100-0-0':

for x in 1 1.5 2 2.5 3.5; do for z in 2000000; do run-nohup -i kml.num-docs.deg$x.$z textgrounder run opennlp.textgrounder.geolocate.GenerateKML -i wotr-may-18-715pm-100-0-0 --kdt num-docs --kml-max-height $z --kml-prefix kml-dist.wotr-may-18-715pm-100-0-0-deg$x.$z --dpc $x &; done; done

This generates KML maps for degree sizes 1, 1.5, 2, 2.5 and 3.5, using
KML max height of 2,000,000, which usually works well.

III. To generate a heatmap of the distribution of a corpus, e.g.
'may-18-715pm-100-0-0':

1. Use WriteGrid to write out the grid to a TextDB database.

for x in 1 1.5 2 2.5 3.5; do textgrounder run opennlp.textgrounder.geolocate.WriteGrid -i wotr-may-18-715pm-100-0-0 --dpc $x -o grid.wotr-may-18-715pm-deg$x; done

2. Use WriteGridToPolygons to convert the TextDB database into R input data
   and copy it to the R working directory (~didir).

for x in 1 1.5 2 2.5 3.5; do textgrounder run opennlp.textgrounder.postprocess.WriteGridToPolygons -i grid.wotr-may-18-715pm-deg$x.data.txt --or wotr-may-18-715pm-deg$x.rect; cp wotr-may-18-715pm-deg$x.rect ~didir; done

Note, you can also output centroids, useful for some R graphs.

for x in 1 1.5 2 2.5 3.5; do textgrounder run opennlp.textgrounder.postprocess.WriteGridToPolygons -i grid.wotr-may-18-715pm-deg$x.data.txt --or wotr-may-18-715pm-deg$x.rect --oc wotr-may-18-715pm-deg$x.centroid; done

You can also output the rectangles and/or centroids in GeoJSON format if needed
using '--geojson'.

3. Edit plot-dist-heatmap.R in ~didir, setting the date appropriately and
   uncommenting the appropriate section for annotated WOTR spans. Open up R
   and execute:

setwd("~/ut/dissertation/dissertation/r-code/")
source("plot-dist-heatmap.R")

This creates a file like 'wotr-may-18-715pm.pdf'.