18 files changed, 103 insertions, 0 deletions
diff --git a/example_geoquery/cdec.ini b/example_geoquery/cdec.ini
new file mode 100644
index 0000000..b9b564d
--- /dev/null
+++ b/example_geoquery/cdec.ini
@@ -0,0 +1,9 @@
+formalism=scfg
+intersection_strategy=cube_pruning
+cubepruning_pop_limit=30
+scfg_max_span_limit=15
+feature_function=WordPenalty
+feature_function=RuleIdentityFeatures
+feature_function=RuleSourceBigramFeatures
+feature_function=RuleTargetBigramFeatures
+feature_function=RuleShape
diff --git a/example_geoquery/cfg.rb b/example_geoquery/cfg.rb
new file mode 100644
index 0000000..94fef2e
--- /dev/null
+++ b/example_geoquery/cfg.rb
@@ -0,0 +1,8 @@
+_PATH          = '/workspace/grounded/test'
+SMT_SEMPARSE   = "python #{_PATH}/smt-semparse/decode_sentence.py /workspace/grounded/test/smt-semparse/work/full_dataset"
+EVAL_PL        = "#{_PATH}/wasp-1.0/data/geo-funql/eval/eval.pl"
+ACCEPT_ZOMBIES = true
+TIMEOUT        = 60
+CDEC_BIN       = '/toolbox/cdec/decoder/cdec'
+$cache         = Memcached.new('localhost:31337')
+
diff --git a/example_geoquery/data.en b/example_geoquery/data.en
new file mode 100644
index 0000000..7d7740e
--- /dev/null
+++ b/example_geoquery/data.en
@@ -0,0 +1,10 @@
+name all the rivers in colorado
+count the states which have elevations lower than what alabama has
+give me the number of rivers in california
+give me the states that border utah
+how high is mount mckinley
+how high is the highest point in america
+how high is the highest point of alabama
+how large is alaska
+how large is texas
+how long is the colorado river
diff --git a/example_geoquery/data.funql b/example_geoquery/data.funql
new file mode 100644
index 0000000..6843394
--- /dev/null
+++ b/example_geoquery/data.funql
@@ -0,0 +1,10 @@
+answer(river(loc_2(stateid('colorado'))))
+answer(count(state(low_point_2(lower_2(low_point_1(stateid('alabama')))))))
+answer(count(river(loc_2(stateid('california')))))
+answer(state(next_to_2(stateid('utah'))))
+answer(elevation_1(placeid('mount mckinley')))
+answer(elevation_1(highest(place(loc_2(countryid('usa'))))))
+answer(elevation_1(highest(place(loc_2(stateid('alabama'))))))
+answer(size(stateid('alaska')))
+answer(size(stateid('texas')))
+answer(len(river(riverid('colorado'))))
diff --git a/example_geoquery/data.gold b/example_geoquery/data.gold
new file mode 100644
index 0000000..aed50af
--- /dev/null
+++ b/example_geoquery/data.gold
@@ -0,0 +1,10 @@
+[riverid(arkansas),riverid(canadian),riverid(colorado),riverid(green),riverid('north platte'),riverid(republican),riverid('rio grande'),riverid('san juan'),riverid('smoky hill'),riverid('south platte')]
+[2]
+[1]
+[stateid(arizona),stateid(colorado),stateid(idaho),stateid(nevada),stateid('new mexico'),stateid(wyoming)]
+[6194]
+[6194]
+[734]
+[591000.0]
+[266807.0]
+[2333]
diff --git a/example_geoquery/data.in b/example_geoquery/data.in
new file mode 100644
index 0000000..e678525
--- /dev/null
+++ b/example_geoquery/data.in
@@ -0,0 +1,10 @@
+<seg grammar="grammar/grammar.3.gz" id="3"> nenne alle flüsse in colorado </seg>
+<seg grammar="grammar/grammar.6.gz" id="6"> zähle die staaten , welche erhebungen haben , die niedriger liegen als jene in alabama </seg>
+<seg grammar="grammar/grammar.15.gz" id="15"> nenne mir die anzahl der flüsse in california </seg>
+<seg grammar="grammar/grammar.16.gz" id="16"> nenne mir die staaten , welche an utah grenzen </seg>
+<seg grammar="grammar/grammar.25.gz" id="25"> wie hoch ist mount mckinley </seg>
+<seg grammar="grammar/grammar.26.gz" id="26"> wie hoch liegt der höchste punkt in amerika </seg>
+<seg grammar="grammar/grammar.29.gz" id="29"> wie hoch liegt der höchste punkt in alabama </seg>
+<seg grammar="grammar/grammar.33.gz" id="33"> wie groß ist alaska </seg>
+<seg grammar="grammar/grammar.34.gz" id="34"> wie groß ist texas </seg>
+<seg grammar="grammar/grammar.37.gz" id="37"> wie lang ist der colorado fluss </seg>
diff --git a/example_geoquery/data.prolog b/example_geoquery/data.prolog
new file mode 100644
index 0000000..6843394
--- /dev/null
+++ b/example_geoquery/data.prolog
@@ -0,0 +1,10 @@
+answer(river(loc_2(stateid('colorado'))))
+answer(count(state(low_point_2(lower_2(low_point_1(stateid('alabama')))))))
+answer(count(river(loc_2(stateid('california')))))
+answer(state(next_to_2(stateid('utah'))))
+answer(elevation_1(placeid('mount mckinley')))
+answer(elevation_1(highest(place(loc_2(countryid('usa'))))))
+answer(elevation_1(highest(place(loc_2(stateid('alabama'))))))
+answer(size(stateid('alaska')))
+answer(size(stateid('texas')))
+answer(len(river(riverid('colorado'))))
diff --git a/example_geoquery/grammar/grammar.15.gz b/example_geoquery/grammar/grammar.15.gz
new file mode 100644
index 0000000..416152f
--- /dev/null
+++ b/example_geoquery/grammar/grammar.15.gz
diff --git a/example_geoquery/grammar/grammar.16.gz b/example_geoquery/grammar/grammar.16.gz
new file mode 100644
index 0000000..19189e6
--- /dev/null
+++ b/example_geoquery/grammar/grammar.16.gz
diff --git a/example_geoquery/grammar/grammar.25.gz b/example_geoquery/grammar/grammar.25.gz
new file mode 100644
index 0000000..a441190
--- /dev/null
+++ b/example_geoquery/grammar/grammar.25.gz
diff --git a/example_geoquery/grammar/grammar.26.gz b/example_geoquery/grammar/grammar.26.gz
new file mode 100644
index 0000000..753e491
--- /dev/null
+++ b/example_geoquery/grammar/grammar.26.gz
diff --git a/example_geoquery/grammar/grammar.29.gz b/example_geoquery/grammar/grammar.29.gz
new file mode 100644
index 0000000..56c91cd
--- /dev/null
+++ b/example_geoquery/grammar/grammar.29.gz
diff --git a/example_geoquery/grammar/grammar.3.gz b/example_geoquery/grammar/grammar.3.gz
new file mode 100644
index 0000000..6e03a98
--- /dev/null
+++ b/example_geoquery/grammar/grammar.3.gz
diff --git a/example_geoquery/grammar/grammar.33.gz b/example_geoquery/grammar/grammar.33.gz
new file mode 100644
index 0000000..4f58842
--- /dev/null
+++ b/example_geoquery/grammar/grammar.33.gz
diff --git a/example_geoquery/grammar/grammar.34.gz b/example_geoquery/grammar/grammar.34.gz
new file mode 100644
index 0000000..a700a54
--- /dev/null
+++ b/example_geoquery/grammar/grammar.34.gz
diff --git a/example_geoquery/grammar/grammar.37.gz b/example_geoquery/grammar/grammar.37.gz
new file mode 100644
index 0000000..32feb04
--- /dev/null
+++ b/example_geoquery/grammar/grammar.37.gz
diff --git a/example_geoquery/grammar/grammar.6.gz b/example_geoquery/grammar/grammar.6.gz
new file mode 100644
index 0000000..bb24c62
--- /dev/null
+++ b/example_geoquery/grammar/grammar.6.gz
diff --git a/example_geoquery/run.sh b/example_geoquery/run.sh
new file mode 100644
index 0000000..249c859
--- /dev/null
+++ b/example_geoquery/run.sh
@@ -0,0 +1,36 @@
+#!/bin/bash
+
+# memcached has to be running!
+#memcached -p 31337
+
+CDEC=/toolbox/cdec
+
+../rebol.rb \
+  -k 100 \
+  -i $(pwd)/data.in \
+  -r $(pwd)/data.en \
+  -g $(pwd)/data.gold \
+  -h $(pwd)/data.funql \
+  -w $(pwd)/../data/weights.init \
+  -t $(pwd)/../data/stopwords.en \
+  -c $(pwd)/cdec.ini \
+  -b $(pwd)/cfg.rb \
+  -o output-weights \
+  -l \
+  -e 0.01 \
+  -j 1 \
+  -u geoquery \
+  -v rebol 2>output.stderr > output.stdout
+
+# translate test
+$CDEC/decoder/cdec \
+  -c cdec.ini \
+  -w output-weights 2>/dev/null \
+  < data.in  \
+  | ../scripts/geoquery/semparse.rb $(pwd)/cfg.rb \
+  | ../scripts/geoquery/query.rb $(pwd)/cfg.rb > output-answers
+
+# evaluate result
+../scripts/geoquery/eval.rb \
+  data.gold < output-answers > output-eval
+