STRI: MeSH

• Description:

  Read in the input MeSH (Medical Subject Headings, MH|SH) and perform ST real-time indexing based on document count. Main Headings and Subheadings are separated by '|'. Also, Subheadings can be represented by two-letter abbreviations or full names.

• Inputs:
  • MeSHs, starred MH and SH are separated by '|'
  • a file, such as 9801.2004.MH.in

• Algorithm:
  • Pre-Processes (Input Filter):
    • Tokenize all Meshs (SH/MH) from the input
    • Filter out illegal Meshs (not in Mh-Jd Table or Sh-Jd Table
    • Assign legal MeSH
  • Processes:
    • Get JD scores for each legal MeSH: jdi
    • Calculate Vector similarity (cosine coefficient) for JDI scores (Word-Jd) and ST-Jd scores.
  • Post-Processes (Output Filter):
    • Print out Input MeSH
    • Print out Legal MeSH
    • Details for output filter
    • Score entries display number
    • No output message
    • ST candidate
    • Cluster option
    • Use alphabetical order for STs have same scores

• Sample commands:

  > stri -imh -p
  => index MeSH from standard input with prompt
  
  > stri -imh -i:9801.2004.MH.in -o:9801.2004.MH.out
  => index MeSH from file, 9801.2004.MH.in, and send results to a file, 9801.2004.MH.out
  

• Sample Outputs:
  • a file, such as 9801.2004.MH.out