Exclusive Filter: A Term Contians a pattern of Parenthetic Acronym

• Description:
  If a term contains a parenthetic acronym, it is not a valid multiword. These terms are filtered out from the MEDLINE n-gram set.

• Examples:
  • the basement membrane zone (BMZ)
  • basement membrane zone (BMZ)
  • membrane zone (BMZ)
  • zone (BMZ)
  • zone (BMZ) of
  • membrane zone (BMZ) of
  • basement membrane zone (BMZ) of
  • zinc finger proteins (ZFPs)

  In this (ACR) parenthetic acronym pattern, it must match the following criteria:

  • Must have left and right parenthesis ( .... )
  • The acronym must be all upper case of A-Z or 0-9
  • The acronym must have more than one character
  • The acronym can't be all digits
  • The last latter of acronym must be the same as the initial letter of it's previous word

  • plural form is also included (ACRs)

• Filter Algorithm:
  • Logics:

    Description	FilterType	Notes
    Get wordList	FT_TBD
    Go through wordList and find ACR and it's previous word	FT_TBD	check left and right parenthesis get ACR from the content between parenthesis get the previous word of (ACR)
    Get ACR singular form from (ACRs)	FT_TBD
    Check if no ACR found (null)	FT_PAR_ACR_NO_PAR	basement membrane zone
    Check if ACR is not uppercase	FT_PAR_ACR_NOT_UPPERCASE	interleukin 12 (p40) hemolymph stimulatory factor (Manduca) regulatory subunit 1 (alpha)
    Check if ACR has only one char	FT_PAR_ACR_SINGLE_CHAR	DeltaPsi (M) antipoly (A) antibody recombinant apolipoprotein (a)
    Check if ACR is all digit	FT_PAR_ACR_ALL_DIGI	yttrium (90) β (2) microglobulin anteiso-C (17:0)
    Check if ACR has no previous word	FT_PAR_ACR_NO_PREV_WORD	(BMZ) of
    Check if last char of ACR is not the same as initial of previosu word	FT_PAR_ACR_NOT_ACR	copper (II) complex T3 (RIA) Plasmodium berghei (ANKA)
    Check if last char of ACR is the same as initial of previosu word	FT_PAR_ACR	membrane zone (BMZ) of zone (BMZ) of

  • source code: FilterParentheticAcronym.java
  • FilterType: FilterType.FT_PAR_ACR

• Accuracy Test on Lexicon:
  • InFile:
    • ${OUT_DATA}/03.LeadEndTerm/lexWords.data
  • Result:

    Lexicon	Filter	Sample No	Pass No	Trap No	Exp No	Pass-Rate
    2023	FT_PAR_ACR	1001867	1001866	1	259	99.9999%
    2022	FT_PAR_ACR	998845	998844	1	259	99.9999%
    2021	FT_PAR_ACR	992545	992544	1	259	99.9999%
    2020	FT_PAR_ACR	983420	983419	1	259	99.9999%
    2019	FT_PAR_ACR	972721	972720	1	258	99.9999%
    2018	FT_PAR_ACR	955564	955563	1	258	99.9999%
    2017	FT_PAR_ACR	935276	935275	1	225	99.9999%
    2016	FT_PAR_ACR	915583	915583	0	204	100.0000%
    2015	FT_PAR_ACR	896213	896213	0	182	100.0000%
    2014	FT_PAR_ACR	875090	875090	0	171	100.0000%