|
1
|
|
|
#!/usr/bin/env python |
|
2
|
|
|
# -*- coding: utf-8 -*- |
|
3
|
|
|
|
|
4
|
|
|
import unittest |
|
5
|
|
|
from processors import * |
|
6
|
|
|
import os |
|
7
|
|
|
|
|
8
|
|
|
|
|
9
|
|
|
__location__ = os.path.realpath(os.path.join(os.getcwd(), os.path.dirname(__file__))) |
|
10
|
|
|
|
|
11
|
|
|
''' |
|
12
|
|
|
Testing named entity recognition. |
|
13
|
|
|
IOB notation should be neutralized for bionlp |
|
14
|
|
|
''' |
|
15
|
|
|
|
|
16
|
|
|
class NERTests(unittest.TestCase): |
|
17
|
|
|
|
|
18
|
|
|
def test_bio_nes(self): |
|
19
|
|
|
json_file = os.path.join(__location__,'serialized_biodoc.json') |
|
20
|
|
|
with open(json_file) as jf: |
|
21
|
|
|
biodoc = Document.load_from_JSON(json.load(jf)) |
|
22
|
|
|
#test .nes for biodoc |
|
23
|
|
|
#print(biodoc.nes) |
|
24
|
|
|
b1_gold_dict = {'TissueType': ['ventral nerve cord', 'neuron', 'nervous system'], 'CellType': ['neurons', 'neurons', 'neurons', 'neurons']} |
|
25
|
|
|
self.assertEqual(b1_gold_dict, biodoc.nes, "document-level nes dict for IOB entities was ill-formed") |
|
26
|
|
|
|
|
27
|
|
|
#test .nes for sentence |
|
28
|
|
|
s = biodoc.sentences[0] |
|
29
|
|
|
#print(s.nes) |
|
30
|
|
|
s1_gold_dict = {'CellType': ['neurons'], 'TissueType': ['ventral nerve cord']} |
|
31
|
|
|
self.assertEqual(s1_gold_dict, s.nes, "sentence-level nes dict for IOB entities was ill-formed") |
|
32
|
|
|
|
|
33
|
|
|
#test non-bio text |
|
34
|
|
|
def test_obama_nes(self): |
|
35
|
|
|
json_file = os.path.join(__location__,'serialized_obama.json') |
|
36
|
|
|
with open(json_file) as jf: |
|
37
|
|
|
doc = Document.load_from_JSON(json.load(jf)) |
|
38
|
|
|
#test .nes for doc |
|
39
|
|
|
#print(doc.nes) |
|
40
|
|
|
d1_gold_dict = {'ORDINAL': ['44th', 'first', 'first'], 'LOCATION': ['US', 'United States', 'United States', 'Honolulu', 'Hawaii', 'Chicago', '13th District', 'United States'], 'NUMBER': ['1', '2', 'three'], 'DATE': ['August 4 , 1961', '1992 and 2004', '1997 to 2004', '2000'], 'ORGANIZATION': ['Columbia University', 'Harvard Law School', 'Harvard Law Review', 'University of Chicago Law School', 'Illinois Senate', 'House of Representatives'], 'MISC': ['American', 'African American', 'Democratic'], 'PERSON': ['Barack Hussein Obama II', 'Obama', 'Bobby Rush']} |
|
41
|
|
|
self.assertEqual(d1_gold_dict, doc.nes, "document-level nes dict for non-IOB entities was ill-formed") |
|
42
|
|
|
#test .nes for sentence |
|
43
|
|
|
s = doc.sentences[0] |
|
44
|
|
|
#print(s.nes) |
|
45
|
|
|
s1_gold_dict = {'ORDINAL': ['44th'], 'DATE': ['August 4 , 1961'], 'NUMBER': ['1', '2'], 'LOCATION': ['US', 'United States'], 'ORGANIZATION': [], 'MISC': ['American'], 'PERSON': ['Barack Hussein Obama II']} |
|
46
|
|
|
self.assertEqual(s1_gold_dict, s.nes, "sentence-level nes dict for non-IOB entities was ill-formed") |
|
47
|
|
|
|
|
48
|
|
|
|
|
49
|
|
|
if __name__ == "__main__": |
|
50
|
|
|
unittest.main() |
myedibleenso /
py-processors
