CPA not consistently lemmatized

[AngledLuffa]

Examples from the dev set (a couple others in train):

# sent_id = email-enronsent23_08-0002
# text = jill allen finishes her cpa today and she and her friends are going to party.
5       cpa     CPA     NOUN    NN      Number=Sing     3       obj     3:obj   _

# sent_id = email-enronsent00_02-0016
# text = As such, we have a CPA, Larry Lewis, working with us to audit and set up transition files.
7       CPA     cpa     NOUN    NN      Number=Sing     5       obj     5:obj|12:nsubj:xsubj    SpaceAfter=No

There's also the Iraq CPA, but that seems orthogonal to the accounting job title

[nschneid]

Thanks! Our English validation script flags a few other potential lemma inconsistencies—do any of these look like they should be fixed? The abbreviations are OK I guess.

! rare lemma America for A/NNP in weblog-blogspot.com_rigorousintuition_20050518101500_ENG_20050518_101500-0070 (majority: A)
! rare lemma central for Central/NNP in reviews-334388-0004 (majority: Central)
! rare lemma McDonald for Mcdonald/NNP in answers-20111019100027AAdxgXV_ans-0005 (majority: Mcdonald)
! rare lemma President for President/NN in newsgroup-groups.google.com_alt.animals_0084bdc731bfc8d8_ENG_20040905_212000-0071, newsgroup-groups.google.com_alt.animals_0084bdc731bfc8d8_ENG_20040905_212000-0166 (majority: president)
! rare lemma Securities for Securities/NNPS in email-enronsent36_01-0030 (majority: Security)
! rare lemma south for South/JJ in reviews-369210-0007 (majority: South)
! rare lemma West for West/JJ in reviews-342807-0001, reviews-342807-0002, reviews-342807-0004 (majority: west)
! rare lemma b for b/NN in email-enronsent44_01-0080, reviews-010433-0003 (majority: benefit)
! rare lemma building for b/NN in email-enronsent35_01-0010 (majority: benefit)
! rare lemma care for c/NN in newsgroup-groups.google.com_eHolistic_2dd76f31ceb6bfe8_ENG_20050513_224200-0056 (majority: c)
! rare lemma class for c/NN in reviews-225632-0001 (majority: c)
! rare lemma Deli for deli/NNP in answers-20111108104228AA6z9uZ_ans-0002 (majority: deli)
! rare lemma glass for glasses/NNS in reviews-363685-0026 (majority: glasses)
! rare lemma Inn for inn/NNP in reviews-159371-0004 (majority: inn)
! rare lemma respects for respects/NNS in answers-20111107144339AA0qw5S_ans-0005 (majority: respect)
! rare lemma science for science/NNP in answers-20111108094740AA1lbom_ans-0003 (majority: Science)
! rare lemma Southwest for southwest/NNP in answers-20111107193044AAvUYBv_ans-0004 (majority: southwest)
! rare lemma versus for vs./IN in email-enronsent27_01-0013, weblog-blogspot.com_rigorousintuition_20060511134300_ENG_20060511_134300-0258 (majority: vs.)

[AngledLuffa]

actually yes, President is sus a couple times

# sent_id = newsgroup-groups.google.com_alt.animals_0084bdc731bfc8d8_ENG_20040905_212000-0071
14      had     have    AUX     VBD     Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin   15      aux     15:aux  _
15      called  call    VERB    VBN     Tense=Past|VerbForm=Part        12      acl:relcl       12:acl:relcl    Cxn=rc-wh-nsubj
16      the     the     DET     DT      Definite=Def|PronType=Art       17      det     17:det  _
17      President       President       NOUN    NN      Number=Sing     15      obj     15:obj|20:nsubj:xsubj   _

# sent_id = newsgroup-groups.google.com_alt.animals_0084bdc731bfc8d8_ENG_20040905_212000-0166
15      whom    whom    PRON    WP      PronType=Rel    18      obj     13:ref  _
16      the     the     DET     DT      Definite=Def|PronType=Art       17      det     17:det  _
17      President       President       NOUN    NN      Number=Sing     18      nsubj   18:nsubj        _
18      called  call    VERB    VBD     Mood=Ind|Number=Sing|Person=3|Tense=Past|VerbForm=Fin   13      acl:relcl       13:acl:relcl    Cxn=rc-wh-obj
19      "       "       PUNCT   ``      _       21      punct   21:punct        SpaceAfter=No
20      Kenny   Kenny   PROPN   NNP     Number=Sing     21      compound        21:compound     _
21      Boy     Boy     PROPN   NNP     Number=Sing     18      xcomp   18:xcomp        SpaceAfter=No
22      "       "       PUNCT   ''      _       21      punct   21:punct        SpaceAfter=No

also vs. is inconsistently annotated vs. or versus

[AngledLuffa]

might i suggest having the script print out line numbers and filenames as well?

[nschneid]

You're welcome to edit https://github.com/UniversalDependencies/UD_English-EWT/blob/dev/not-to-release/tools/neaten.py—look for lemma_docs