Generating Synonyms from Lexicon Nominalizations

I. Introduction

Nominalizations are synonyms. In order to save the linguist's tagging time, all nominalization pairs (nPairs) are removed from the the sClass candididate list (so the linguists does not need to tag them and then they are added back automatically). Please see the following two examples:

• Example-1 (this type of sClass are taken care of in synonymMeta.data):
  • sClass
    #SYNONYM_CLASS|C0001807|Aggressive behavior
128|E0007791|aggression|Y
128|E0007793|aggressiveness|Y
128|E0528674|aggressive|Y

  • nominalizations:
    • 1024|E0221219|aggress (nom of 128|E0007791|aggression)
    • 1|E0007792|aggressive (nom of 128|E0007793|aggressiveness)

    • Outputs: All five of above synonyms are used to generate sPairs
  • Example-2 (this type of sClass are taken care of in synonymNom.data):
    • sClass: no E0001532|Christian in sClass

    • Outputs: the nominalizations (synonyms) below are used to generate a sPair.
      • Christian|adj|E0001532
      • Christianity|noun|E0001534

  II. Process

  • Directory: ${LEXICON_SYNONYMS}
  • program: ./Synonym/GenSynonymFromNom.java
  • Script:
    • shell>cd ./bin/
    • shell>GetSynonyms ${YEAR} > log.14
14
  • Inputs:
    • ./inData/LRNOM
    • ./outData/Results/synonymFromMeta.data (copy from synonymFromMeta.data.${YEAR})
  • Algorithm:
    • Go though all nominalizations to generate sPairs
    • Exclude those are already in synonymMeta
      => The number could be decreased if the sPairs from CUI covers more number than new increase nomalizationin the Lexicon.
    • print to output file
  • Outputs:
    • ./outData/Results/synonymFromNom.data
    • Format:

      NpLc Synonym-1

      Synonym-1

      Pos-1

      Synonym-2

      Pos-2

      EUI-1