Commits

Author Commit Message Labels Comments Date
Ben Wing
Look for German-style 72/45/30/N coordinates in the degree field whenever converting separate deg/min/sec specifiers, since we have cases where 'breitengrad' has a German-style spec, but also cases where both 'breitengrad' and 'breitenminute' occur
Ben Wing
Add support for stopniN/stopniW/etc. (built-in lat/long); also handle breitengrad, etc. differently since breitenminute etc. do exist; add more language-specific argument names
Ben Wing
manual merge
Ben Wing
Implement Dirichlet smoothing properly, add a few aliases, etc.
Ben Wing
Major rewrite of code to make it easier to incorporate different smoothing algorithms and ways of filtering the distributions
Ben Wing
Add code to implement mean shift
Mike Speriosu
Added KNNKMLGenerator.
Mike Speriosu
Added command line option to run on geotwitterut-small.
Mike Speriosu
First code in place to look at kNN for exploration purposes.
Ben Wing
Automatic merge
Ben Wing
Fix so that output goes to stdout when necessary even when a split is happening; also bound long strings in warning messages to a max length
Ben Wing
Need to mark strings with non-ASCII chars as Unicode
Ben Wing
Fix problems with no output to stdout -- output all output to stderr (to make it clearer what's going on, so stderr/stdout synched) only when debugging turned on
Ben Wing
Add quick-start about run-preprocess
Ben Wing
Add support for German-language wiki geocoordinates
Ben Wing
processwiki.py: Fix bugs that cause crashes parsing refs; add support for Portuguese geotags
Mike Speriosu
Improved PreprocWikiDump to find more geotag types.
Mike Speriosu
Added first version of PreprocWikiDump.
Ben Wing
Move all the old, no-longer-used stuff into src/old (a bit of it was in fact used)
Ben Wing
Fix compile problem
Ben Wing
More conservative imports of gridlocate
Ben Wing
More conservative imports of worddist
Ben Wing
Move files into gridlocate and worddist packages
Ben Wing
Need to resurrect the non-parallel way for now; you get an assertion failure after 1200 or so documents in GeoText otherwise, not clear why
Ben Wing
manual merge; at least it compiles, not tested
Ben Wing
manual merge
Ben Wing
Fix bug that led to huge (10-15%) decrease in accuracy across the board; mysterious why the bug fix works, possibly a Scala bug
Ben Wing
Print out more info about KL-div, number of types/tokens
Ben Wing
Bug fix that formerly led to error crash
Ben Wing
Add some comments about GridLocate
  1. Prev
  2. Next