add desambiguation evaluation script

.gitignore

@@ -147,4 +147,5 @@ notes.md
 .idea*
 other/*
 test*
-nohup.out
+nohup.out
\ No newline at end of file
+log*
\ No newline at end of file

combination_embeddings.py

@@ -136,7 +136,6 @@ meta_data = MetaDataSerializer(
 )
 meta_data.save("outputs/{0}.json".format(PREFIX_OUTPUT_FN))
 
-
 #############################################################################################
 ################################# LOAD DATA #################################################
 #############################################################################################

eval_amb.py

+from glob import glob
+import os
+os.environ['CUDA_VISIBLE_DEVICES'] = '-1' # No need for GPU
+import argparse
+import logging
+
+import pandas as pd
+
+from predict_toponym_coordinates import Geocoder
+from lib.geo import haversine_pd
+
+logging.getLogger("tensorflow").setLevel(logging.CRITICAL)
+logging.getLogger("tensorflow_hub").setLevel(logging.CRITICAL)
+
+
+parser = argparse.ArgumentParser()
+parser.add_argument("eval_dataset")
+parser.add_argument("models_directory")
+args = parser.parse_args()
+
+
+
+EVAL_DATASET_FN= args.eval_dataset#"./test_dataset_ambiguity.csv"
+
+
+def eval_model(eval_dataset_fn,model_fn,model_index_fn):
+    df = pd.read_csv(eval_dataset_fn,index_col=0)
+    geocoder = Geocoder(model_fn,model_index_fn)
+    lon,lat = geocoder.get_coords(df.name1.values,df.name2.values)                  
+    lon,lat = geocoder.wgs_coord(lon,lat)
+
+    df["p_longitude"] = lon
+    df["p_latitude"] = lat
+
+    df["dist"] = haversine_pd(df.longitude,df.latitude,df.p_longitude,df.p_latitude)
+
+    print("Dataset -- {0} -- Model -- {1}".format(\
+        eval_dataset_fn.split("/")[-1],
+        model_fn.split("/")[-1]))
+    print("100km",(df.dist<100).sum()/len(df))
+    print("50km",(df.dist<50).sum()/len(df))
+    print("20km",(df.dist<20).sum()/len(df))
+
+prefixes = [x.rstrip(".h5") for x in glob(args.models_directory+"/*.h5")]
+for prefix in prefixes:
+    eval_model(EVAL_DATASET_FN,prefix + ".h5",prefix + "_index")
\ No newline at end of file

predict_toponym_coordinates.py

 from keras.models import load_model
+import os
 import tensorflow as tf
 import keras.backend as K
-from utils import NgramIndex
+from lib.ngram_index import NgramIndex
 import numpy as np
 
 from tensorflow.python.keras.backend import set_session
 from tensorflow.python.keras.models import load_model
 
+
 sess = None
 graph = None
 
@@ -75,6 +77,16 @@ class Geocoder(object):
         #     with graph.as_default():
         lon,lat = self.keras_model.predict([[p],[c]])
         return lon[0][0],lat[0][0]
+    
+    def get_coords(self,list_toponym,list_toponym_context):
+        p = [self.ngram_encoder.complete(self.ngram_encoder.encode(toponym),self.ngram_encoder.max_len) for toponym in list_toponym]
+        c = [self.ngram_encoder.complete(self.ngram_encoder.encode(toponym),self.ngram_encoder.max_len) for toponym in list_toponym_context]
+
+        p = np.array(p)
+        c = np.array(c)
+        
+        lon,lat = self.keras_model.predict([p,c])
+        return lon,lat
 
     def wgs_coord(self,lon,lat):
         return ((lon*360)-180),((lat*180)-90)
@@ -98,11 +110,7 @@ class Geocoder(object):
             ax.plot(lon,lat,marker='o', color='red', markersize=5)
             plt.show()
 
-geocoder = Geocoder("outputs/LSTM_FR.txt_100_4_0.002_None_A_I_C.h5","./outputs/FR.txt_100_4_0.002_None_A_I_C_index")
-top,topc = "Paris","Cherbourg"
-lon,lat = geocoder.get_coord(top,topc)
-lon,lat = geocoder.wgs_coord(lon,lat)
-geocoder.plot_coord("{0},{1}".format(top,topc),lat,lon)
+
 
 if __name__ == "__main__":
     from flask import Flask, escape, request, render_template

run_train.py

 from lib.run import GridSearchModel
 from collections import OrderedDict
 
+# Build all combination of relations
 rels = ["-i","-a","-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_US_FR.txt"]
-comb = []
+combinations = []
 for rel in rels:
-    comb.append(rel)
+    combinations.append(rel)
     for rel2 in rels:
         if not rel == rel2:
-            if not rel2+ " " + rel in comb:
-                comb.append(rel+ " " + rel2)
+            if not rel2+ " " + rel in combinations:
+                combinations.append(rel+ " " + rel2)
 
+# Init GridsearchModel
 grid = GridSearchModel(\
     "python3 combination_embeddings.py",
-    **OrderedDict({
-    "rel":['-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_US_FR.txt','-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_US_FR.txt -i', '-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_US_FR.txt -a','-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_US_FR.txt -a -i'],#[comb],
+    **OrderedDict({ # necessary because some args have to be given in a certain order
+    "rel":combinations,
     "-n":[4],
-    "--ngram-word2vec-iter" :[1],
+    "--ngram-word2vec-iter" :[50],
     "-e":[100],
-    "geoname_fn":"../data/geonamesData/US_FR.txt".split(),
-    "hierarchy_fn":"../data/geonamesData/hierarchy.txt".split(),
-    "store_true":["rel"]
+    "geoname_fn":"../data/geonamesData/FR.txt".split(),
+    "hierarchy_fn":"../data/geonamesData/hierarchy.txt".split()
     }.items()))
+
 print("########### THE FOLLOWING COMMAND(S) WILL BE EXECUTED ###########" )
 [print(task.get_command()) for task in grid.tasks]
 print("#################################################################")
-grid.run("log_RUN_TEXAS_IDFrance.txt")
+grid.run("outputs/log_RUN_TEXAS_IDFrance.txt")
\ No newline at end of file
+
+#["-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_FR.txt","-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_FR.txt -a","-w --wikipedia-cooc-fn ../data/wikipedia/cooccurrence_FR.txt -i"]
\ No newline at end of file