adding neighborhoods, probs to geostring return

lib/tagnews/geoloc/tag.py

@@ -291,22 +291,26 @@ def extract_geostrings(self, s, prob_thresh=0.5):
         geostrings : list of lists of strings
             The list of extracted geostrings from the article text.
             Each word is kept separated in the list.
-            Examle:
+            Example:
                 [['1300', 'W.', 'Halsted'], ['Ohio']]
         """
         words, probs = self.extract_geostring_probs(s)
         above_thresh = probs >= prob_thresh
 
         words = ["filler"] + words + ["filler"]
+        probs = np.append(0, np.append(probs, 0))
+
         above_thresh = np.concatenate([[False], above_thresh, [False]]).astype(np.int32)
         switch_ons = np.where(np.diff(above_thresh) == 1)[0] + 1
         switch_offs = np.where(np.diff(above_thresh) == -1)[0] + 1
 
         geostrings = []
+        probstrings = []
         for on, off in zip(switch_ons, switch_offs):
             geostrings.append(words[on:off])
+            probstrings.append(probs[on:off])
 
-        return geostrings
+        return geostrings, probstrings
 
     @staticmethod
     def lat_longs_from_geostring_lists(geostring_lists, **kwargs):

lib/tagnews/utils/neighborhoods.py

@@ -0,0 +1,97 @@
+neighborhoods = [
+    "Andersonville",
+    "Archer Heights",
+    "Ashburn",
+    "Ashburn Estates",
+    "Austin",
+    "Avaondale",
+    "Belmont Central",
+    "Beverly",
+    "Beverly Woods",
+    "Brainerd",
+    "Bridgeport",
+    "Brighton Park",
+    "Bronceville",
+    "Bucktown",
+    "Burnside",
+    "Calumet Heights",
+    "Canaryville",
+    "Clearing",
+    "Chatham",
+    "Chinatown",
+    "Cottage Grove Heights",
+    "Cragin",
+    "Dunning",
+    "East Chicago",
+    "Edison Park",
+    "Edgebrook",
+    "Edgewater",
+    "Englewood",
+    "Ford City",
+    "Gage Park",
+    "Galewood",
+    "Garfield Park",
+    "Garfield Ridge",
+    "Gold Coast",
+    "Grand Crossing",
+    "Gresham",
+    "Hamilton Park",
+    "Humboldt Park",
+    "Hyde Park",
+    "Jefferson Park",
+    "Kelvyn Park",
+    "Kenwood",
+    "Kilbourn Park",
+    "Lake Meadows",
+    "Lakeview",
+    "Lawndale",
+    "Lincoln Park",
+    "Lincoln Square",
+    "Little Village",
+    "Logan Square",
+    "Longwood Manor",
+    "Loop",
+    "Marquette Park",
+    "McKinley Park",
+    "Midway",
+    "Morgan Park",
+    "Montclare",
+    "Mount Greenwood",
+    "North Center",
+    "Norwood Park",
+    "Old Irving Park",
+    "Old Town",
+    "Park Manor",
+    "Pilsen",
+    "Princeton Park",
+    "Portage Park",
+    "Pullman",
+    "Ravenswood",
+    "River North",
+    "River West",
+    "Rodgers Park",
+    "Roscoe VIllage",
+    "Roseland",
+    "Sauganash",
+    "Schorsch Village",
+    "Scottsdale",
+    "South Chicago",
+    "South Deering",
+    "South Loop",
+    "South Shore",
+    "Streeterville",
+    "Tri-Taylor",
+    "Ukrainian Village",
+    "United Center",
+    "Uptown",
+    "Vittum Park",
+    "Washington Heights",
+    "West Elsdon",
+    "West Loop",
+    "West Pullman",
+    "Westlawn",
+    "Wicker Park",
+    "Woodlawn",
+    "Wrigleyville",
+    "Wrigtwood",
+]