change.py

src/change.py

@@ -0,0 +1,45 @@
+#!/usr/bin/env python3
+
+"""
+./change.py --corpus <filename> --output_vectors <filename> --output_model <filename>
+"""
+
+import gensim
+import argparse
+from scipy import spatial
+import csv
+
+parser = argparse.ArgumentParser()
+parser.add_argument("--inputPrefix")
+parser.add_argument("--words")
+parser.add_argument("--year_points")
+parser.add_argument("--output_csv")
+#parser.add_argument("--model")
+args = parser.parse_args()
+print args
+args.words = args.words.split('_')
+args.year_points = args.year_points.split('_')
+print args.words
+print type(args.year_points)
+
+# Define measures of interest
+word2itself = dict()
+
+for idx, y in enumerate(args.year_points):
+    # load the model of each year
+    model = gensim.models.Word2Vec.load(args.inputPrefix + y)
+    # define reference year
+    if y=='1750':
+        model_ref = model
+    # compute analysis of interest per word
+    for w in args.words:
+#        print model[y][w]        
+#        word2itself[w].append()
+#        print model[w].transpose()
+
+        print [w, spatial.distance.cosine(model[w].transpose(),model[w].transpose()), spatial.distance.cosine(model_ref[w].transpose(),model[w].transpose())]
+
+#f = open(args.output_csv,'wb')
+#writing = csv.DictWriter(f,word2itself.keys())
+#writing.writerows(word2itself)
+#f.close()