Add out.tsv

dev-0/out.tsv

@@ -0,0 +1,83 @@
+effective_date=2014-05-20 jurisdiction=New_York 
+effective_date=2012-06-21 jurisdiction=Georgia party=An_Electric_Membership_Corporation 
+jurisdiction=Delaware term=2_years
+effective_date=2007-01-23 jurisdiction=Massachusetts 
+jurisdiction=Delaware 
+jurisdiction=Washington party=Schneider 
+jurisdiction=Delaware 
+jurisdiction=New_Hampshire party=Nat_Krishnamurti party=Interpace_Diagnostics_Group party=Interpace_Diagnostics_Corporation 
+jurisdiction=New_York 
+effective_date=2015-11-11 
+""
+jurisdiction=Delaware 
+jurisdiction=Delaware 
+effective_date=2000-05-23 jurisdiction=Delaware 
+jurisdiction=New_York 
+jurisdiction=Massachusetts 
+effective_date=2008-07-31 jurisdiction=California 
+jurisdiction=California 
+jurisdiction=New_York 
+effective_date=2017-07-11 jurisdiction=California 
+""
+jurisdiction=California 
+jurisdiction=Virginia party=The_Corporation 
+""
+""
+jurisdiction=Indiana 
+jurisdiction=New_Jersey 
+jurisdiction=California 
+jurisdiction=Delaware term=2_years
+jurisdiction=Georgia 
+jurisdiction=Delaware 
+effective_date=2015-06-23 jurisdiction=New_York 
+jurisdiction=California 
+""
+jurisdiction=California 
+""
+party=Target_Corporation party=Sears_Holdings_Corporation 
+jurisdiction=Ohio 
+jurisdiction=Massachusetts 
+jurisdiction=California 
+effective_date=2000-02-10 
+jurisdiction=California 
+jurisdiction=Delaware 
+jurisdiction=Illinois 
+""
+effective_date=2010-06-23 jurisdiction=Texas 
+jurisdiction=New_Jersey 
+jurisdiction=Delaware party=LaserCard_Corporation 
+effective_date=2001-01-26 jurisdiction=Washington 
+jurisdiction=Georgia 
+effective_date=2014-11-26 jurisdiction=Delaware 
+jurisdiction=Oregon 
+jurisdiction=California 
+effective_date=2012-06-11 jurisdiction=Delaware 
+jurisdiction=Delaware 
+jurisdiction=Massachusetts 
+jurisdiction=California party=Life\nTechnologies_Corporation party=Invitrogen_Corporation 
+effective_date=2012-09-20 party=Contran_Corporation 
+jurisdiction=Delaware 
+effective_date=2008-09-15 jurisdiction=Illinois party=Paul_B._Mulhollem party=Paul_B._Mulhollem\nPAUL_B._MULHOLLEM\nDated:_September_29 
+""
+jurisdiction=Ohio term=2_years
+jurisdiction=California 
+jurisdiction=New_York 
+jurisdiction=California 
+effective_date=2015-03-16 jurisdiction=Utah 
+""
+jurisdiction=Washington 
+jurisdiction=Virginia 
+jurisdiction=New_York 
+jurisdiction=New_York party=Eurand_N.V. 
+jurisdiction=Illinois 
+jurisdiction=California 
+""
+""
+effective_date=2011-03-29 jurisdiction=Texas term=2_years
+effective_date=2011-05-26 jurisdiction=California 
+jurisdiction=Washington party=Digital_River 
+jurisdiction=Delaware 
+jurisdiction=Washington 
+effective_date=2011-01-18 jurisdiction=New_York term=2_years
+jurisdiction=Washington party=James_B._Bucher 
+jurisdiction=Illinois 

main.py

@@ -114,28 +114,29 @@ def getTerm(document):
 
 
 if __name__ == '__main__':
-    NDAs = readInput('train/in.tsv.xz')
+    NDAs = readInput('test-A/in.tsv.xz')
 
     ner = spacy.load('NER')
 
-    predicted = [''] * len(NDAs)
+    predicted = []
-
-    document = ner(NDAs[9])
 
     for i in range(len(NDAs)):
-        document = ner(NDAs[i])
+        document = ner(NDAs[i].replace('\n', ' '))
+        predict = ''
 
         ed = getEffectiveDate(document)
         j = getJurisdiction(document)
         p = getParties(document)
         t = getTerm(document)
 
-        if len(ed) > 0: predicted[i] += 'effective_date=' + ed + ' '
+        if len(ed) > 0: predict += 'effective_date=' + ed + ' '
-        if len(j) > 0: predicted[i] += 'jurisdiction=' + j + ' '
+        if len(j) > 0: predict += 'jurisdiction=' + j + ' '
         if len(p) > 0:
-            for party in p: predicted[i] += 'party=' + party + ' '
+            for party in p: predict += 'party=' + party + ' '
-        if len(t) > 0: predicted[i] += 'term=' + t
+        if len(t) > 0: predict += 'term=' + t
 
-    with open('train/out.tsv', 'w', newline='') as f:
+        predicted.append([predict])
+
+    with open('test-A/out.tsv', 'w', newline='') as f:
         writer = csv.writer(f)
         writer.writerows(predicted)

test-A/out.tsv

@@ -0,0 +1,203 @@
+jurisdiction=Ohio 
+jurisdiction=California 
+jurisdiction=Indiana 
+effective_date=2007-03-28 jurisdiction=New_York term=2_years
+jurisdiction=Nevada 
+effective_date=2010-07-16 jurisdiction=Utah 
+""
+jurisdiction=California 
+jurisdiction=Delaware 
+effective_date=2008-06-25 jurisdiction=New_York 
+jurisdiction=California 
+jurisdiction=Delaware term=3_years
+jurisdiction=New_York party=Central_European_Distribution_Corporation 
+effective_date=2017-07-20 jurisdiction=Georgia 
+jurisdiction=Texas party=Open_Text_Corporation 
+effective_date=2008-04-16 jurisdiction=California 
+"jurisdiction=Washington party=L._Schneider, party=Schneider "
+""
+jurisdiction=Florida 
+""
+jurisdiction=Florida 
+jurisdiction=New_York 
+jurisdiction=California 
+jurisdiction=Florida 
+jurisdiction=New_York 
+jurisdiction=Washington party=ELECTRONIC_ARTS 
+jurisdiction=Texas 
+jurisdiction=Delaware 
+jurisdiction=Georgia party=American_Tower_Corporation 
+jurisdiction=Washington 
+effective_date=2012-01-11 jurisdiction=Delaware 
+""
+jurisdiction=California party=Rovi_Corporation party=Sonic_Solutions 
+jurisdiction=California 
+""
+jurisdiction=California party=CVS_Caremark_Corporation 
+jurisdiction=Virginia 
+effective_date=2014-01-28 jurisdiction=Delaware 
+jurisdiction=Florida 
+jurisdiction=California 
+jurisdiction=Virginia 
+effective_date=2010-07-23 jurisdiction=Pennsylvania 
+effective_date=2010-01-25 jurisdiction=Delaware 
+jurisdiction=Washington 
+jurisdiction=Virginia 
+jurisdiction=California 
+""
+effective_date=2000-03-28 jurisdiction=Pennsylvania 
+jurisdiction=New_Jersey 
+jurisdiction=Illinois 
+""
+jurisdiction=Massachusetts 
+jurisdiction=Washington 
+jurisdiction=New_York 
+""
+jurisdiction=Pennsylvania 
+jurisdiction=Pennsylvania 
+jurisdiction=California 
+effective_date=2001-09-17 jurisdiction=Georgia 
+jurisdiction=Delaware term=2_years
+effective_date=2018-04-10 jurisdiction=Delaware 
+jurisdiction=Massachusetts 
+jurisdiction=Connecticut 
+effective_date=2006-05-11 jurisdiction=California 
+jurisdiction=Texas 
+jurisdiction=Delaware 
+jurisdiction=Virginia party=Peter_Federico 
+effective_date=2013-11-18 
+jurisdiction=Washington 
+""
+party=The_Corporation 
+jurisdiction=Texas 
+""
+""
+effective_date=2017-05-24 jurisdiction=Nevada 
+jurisdiction=New_York 
+jurisdiction=Delaware party=Sprint_Corporation 
+jurisdiction=Illinois 
+effective_date=2014-12-22 jurisdiction=New_York 
+effective_date=2012-01-27 jurisdiction=New_York term=18_months
+jurisdiction=Florida 
+jurisdiction=Virginia party=L._Waechter_CACI_International 
+jurisdiction=New_York 
+effective_date=2005-09-19 jurisdiction=Nevada party=MIKOHN_GAMING_CORPORATION 
+effective_date=2016-11-17 jurisdiction=Delaware 
+jurisdiction=New_Jersey 
+""
+jurisdiction=Georgia party=EFCO_Corporation 
+""
+jurisdiction=California party=Laboratory_Corporation_of_America_Holdings party=Laboratory_Corporation 
+jurisdiction=Missouri 
+jurisdiction=Florida 
+effective_date=2015-08-26 jurisdiction=Washington 
+jurisdiction=Ohio 
+effective_date=2004-03-17 jurisdiction=New_York 
+jurisdiction=Georgia 
+jurisdiction=New_York 
+jurisdiction=New_York 
+""
+jurisdiction=California 
+party=Telco_Solutions 
+jurisdiction=Ohio 
+""
+jurisdiction=California 
+jurisdiction=California 
+jurisdiction=Connecticut 
+effective_date=2019-02-14 
+effective_date=2016-11-17 jurisdiction=Delaware 
+effective_date=2019-01-16 jurisdiction=Utah 
+jurisdiction=California term=18_months
+jurisdiction=Utah 
+""
+jurisdiction=New_York 
+jurisdiction=New_York 
+effective_date=2002-02-28 jurisdiction=California party=SignalSoft_Corporation 
+effective_date=2013-07-29 jurisdiction=New_York 
+effective_date=2006-04-21 jurisdiction=Virginia 
+jurisdiction=Massachusetts 
+effective_date=2006-07-18 jurisdiction=Texas term=2_years
+jurisdiction=Massachusetts 
+effective_date=2007-04-23 
+effective_date=2018-04-26 jurisdiction=New_York 
+party=Flagstar_Bancorp 
+jurisdiction=Pennsylvania 
+jurisdiction=New_York 
+jurisdiction=New_Jersey 
+effective_date=2011-05-27 jurisdiction=Texas 
+""
+jurisdiction=Ohio 
+jurisdiction=Oregon 
+effective_date=2014-07-15 jurisdiction=New_Jersey party=Realogy_Group_LLC 
+effective_date=1999-04-29 jurisdiction=Florida 
+jurisdiction=Oregon 
+jurisdiction=Massachusetts 
+jurisdiction=Ohio 
+jurisdiction=California term=18_months
+""
+jurisdiction=Indiana 
+jurisdiction=California 
+effective_date=2012-05-21 jurisdiction=Virginia party=Donald_H._Layton\nDate:\nMay 
+""
+effective_date=2009-02-16 jurisdiction=Delaware 
+party=2212421\n(Federal_ID# 
+jurisdiction=Pennsylvania party=Ikonics_Corporation 
+jurisdiction=Ohio 
+effective_date=1994-07-11 jurisdiction=California 
+jurisdiction=New_Jersey 
+""
+""
+jurisdiction=Virginia party=The_Corporation 
+""
+""
+jurisdiction=Delaware party=Chugai_Pharmaceutical_Co. 
+effective_date=2006-02-28 jurisdiction=New_York 
+jurisdiction=California 
+""
+effective_date=2009-01-15 jurisdiction=Texas party=1st 
+effective_date=2007-11-30 jurisdiction=Georgia 
+effective_date=1998-09-22 jurisdiction=New_York 
+effective_date=2012-01-27 jurisdiction=New_York term=18_months
+jurisdiction=Oregon 
+jurisdiction=Nevada 
+jurisdiction=California 
+""
+effective_date=2008-03-12 jurisdiction=California 
+jurisdiction=Georgia party=Innotrac_Corporation 
+""
+jurisdiction=New_York 
+""
+jurisdiction=New_York party=General_Electric_Company 
+jurisdiction=Indiana 
+jurisdiction=Georgia 
+jurisdiction=New_Jersey 
+""
+party=The_Cato_Corporation 
+jurisdiction=Georgia term=2_years
+""
+effective_date=2019-07-19 jurisdiction=Nevada party=Cosmos_Group_Holdings 
+jurisdiction=Missouri 
+effective_date=2007-06-14 jurisdiction=Utah 
+effective_date=2010-04-27 jurisdiction=Connecticut 
+jurisdiction=Delaware 
+effective_date=2015-11-24 
+effective_date=2017-02-10 
+jurisdiction=Georgia 
+jurisdiction=California 
+effective_date=2013-05-29 jurisdiction=New_York party=WMI_Holdings_Corp. 
+jurisdiction=Pennsylvania 
+jurisdiction=California party=SPECIALIZED_MARKETING_SERVICES 
+""
+jurisdiction=California 
+jurisdiction=Nevada party=IPSA_International_Services 
+jurisdiction=Pennsylvania term=2_years
+effective_date=2003-07-30 jurisdiction=New_York party=Dodger_Acquisition_Corp. 
+jurisdiction=Virginia party=Daniel_L._Betts\nBY 
+jurisdiction=Virginia 
+jurisdiction=Delaware 
+jurisdiction=Nevada 
+jurisdiction=Nevada 
+jurisdiction=Georgia 
+effective_date=2008-09-10 jurisdiction=California 
+""
+""