Loading ...
Sorry, an error occurred while loading the content.

Skimmer Callsign Validation

Expand Messages
  • Pete Smith
    One of the early criticisms of the first version of CW Skimmer was that it posted too many busted calls. In response to that criticism, Alex has introduced,
    Message 1 of 1 , Jun 29, 2008
    • 0 Attachment
      One of the early criticisms of the first version of CW Skimmer was that it
      posted too many busted calls. In response to that criticism, Alex has
      introduced, in version 1.2, a user-selectable 4-level validation
      scheme, ranging from "minimal" to "paranoid." Each level tests the
      callsign as copied against progressively tougher criteria, both in terms of
      the number of times it must be copied and in whether it matches against
      patterns derived from an analysis of 11 years of cluster spots. The
      "paranoid" level does not post a callsign unless it matches one in a
      master.dta file, known among contesters as "Super Check Partial."

      During the recent WPX CW contest, DL6MHW recorded 17 minutes on 40 meter
      CW. Using his recording, I ran CW Skimmer in each of the four validation
      modes, told it only to spot stations that were CQ-ing, and collected the
      spots. WG4M, a Microsoft Access maven, analyzed the lists of spots and
      arranged them in a table that shows each unique callsign copied by Skimmer
      and how it fared through the four validation modes. You can clearly see
      how Skimmer now becomes progressively more "skeptical" when the more
      stringent validation modes are used.

      There is a Microsoft Excel file at
      www.pvrc.org/~n4zr/Summary_of_All_Calls_6-27-08.xls that summarizes the
      results. Callsigns color-coded red in the left-most column were not found
      in QRZ.com but were accepted by one or more of the validation levels above
      minimal - at least some may be good, because QRZ is not perfectly timely or
      inclusive. Similarly, call-signs colored yellow in that column were found
      in QRZ but were not validated by any level above minimal (I call them false
      negatives).

      The results aren't perfect - but I think they clearly show that Skimmer is
      now at least comparable in accuracy to the average human op who posts spots
      on the DX cluster network. Not bad! If you want to see for yourself,
      click on the URL above - if you enter it manually, be sure you put
      underscore characters everywhere there appears to be a space. Comments
      welcome!


      73, Pete N4ZR
    Your message has been successfully submitted and would be delivered to recipients shortly.