Jump to content


Photo

[COUNT ?] Question


  • Please log in to reply
9 replies to this topic

#1 Randy Steffens Jr

Randy Steffens Jr

    Silver

  • Active Members
  • PipPipPip
  • 173 posts
  • Accordance Version:10.x

Posted 30 May 2013 - 09:31 AM

First is there a simple way to tell the [COUNT] command to count everything higher or lower than a certain number? The equivalent of this: [COUNT >1]?

 

When I search the KJV Bible for every word that occurs in the entire text by doing a search for * (asterisk), I get a total of 13760 distinct word occurrences in analytics.  When I attempt to do the same search using [COUNT 1-100000] (100000 being a randomly chosen, ridiculously high number to include everything), I get the number 13754 in analytics.  Six very common words, that are mentioned several thousand times in the KJV, are left out, like "be", "were", and "from".  Why is this, and how can I correct it?

 

Randy

 

 



#2 James Tucker

James Tucker

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 637 posts
  • Gender:Not Telling
  • Accordance Version:10.x

Posted 30 May 2013 - 11:11 AM

Looks like a bug to me



#3 Steve King

Steve King

    Gold

  • Active Members
  • PipPipPipPip
  • 366 posts
  • Gender:Male
  • Location:Hoddesdon, Hertfordshire, UK
  • Accordance Version:10.x

Posted 30 May 2013 - 01:35 PM

ESV has a similar problem but only misses one word 'be' with a lowercase 'b'. 'Be' is found OK.



#4 James Tucker

James Tucker

    Platinum

  • Active Members
  • PipPipPipPipPip
  • 637 posts
  • Gender:Not Telling
  • Accordance Version:10.x

Posted 30 May 2013 - 02:02 PM

Yes, somewhere in the search algorithm cases such as this (e.g., be -> Be, THY - Thy, We -> we, Were -> were) are discarded and not counted, when in fact they should either be included under one lexical form (e.g., Be (2) + be (50) = be (52)), or listed separately with each count under the respective form. Hence, whatever data set you use will not eradicate the issue—it will only tell you (1) that the algorithm isn't counting correctly, and (2) that there are x amount of inflected forms not considered.



#5 Timothy Jenney

Timothy Jenney

    Platinum

  • Accordance
  • 1,436 posts
  • Gender:Male
  • Location:sunny Winter Haven, FL
  • Interests:a good cup of coffee, sci-fi, playing bass, listing to jazz and the blues, camping, fishing and the great outdoors
  • Accordance Version:10.x

Posted 30 May 2013 - 03:51 PM

I am not sure of the answer to your question, but I did try to replicate the issue. I used Accordance 10.1.6 and the KJVS text version 3.2.

 

I ran the search two ways. First, I simply searched for all words using the asterisk (*) and opened an analysis pane. According to this search, there are 13,760 forms in the KJV text.

 

I then ran the search using the COUNT command, but increased the number to ten million (just in case! ;-)): [COUNT 1-10,000,000]. According to this search, there are 13754 forms in the text. That's a difference of six forms. In fact, one can see the difference in the search results [attached screenshots], as the words not counted are still in black in the Search Results windows.

 

So, issue confirmed, though I can't explain it. I imagine some one more savvy than I will pop up here eventually and let us know what the difference is.

Attached Files


Edited by Timothy Jenney, 30 May 2013 - 04:01 PM.

Blessings,
"Dr. J"

Timothy P. Jenney, Ph. D.
"Lighting the Lamp" Host and Producer

#6 Randy Steffens Jr

Randy Steffens Jr

    Silver

  • Active Members
  • PipPipPip
  • 173 posts
  • Accordance Version:10.x

Posted 31 May 2013 - 07:36 AM

Thanks for your input. After further analysis, it appear that these particular words are never counted in ANY [COUNT] search, no matter what arguments are used with the COUNT command. For example a search for the word "were" shows it appears 2768 times in the KJV. However a [COUNT 2768] search turns up with nothing.

#7 Helen Brown

Helen Brown

    Mithril

  • Admin
  • 7,988 posts
  • Gender:Female
  • Location:heart in Israel
  • Accordance Version:10.x

Posted 31 May 2013 - 08:57 AM

We are looking into this issue, and it seems to reside in the compilation of the text rather than in Accordance. We are working on it now.


Helen Brown
OakTree Software

#8 Randy Steffens Jr

Randy Steffens Jr

    Silver

  • Active Members
  • PipPipPip
  • 173 posts
  • Accordance Version:10.x

Posted 31 May 2013 - 10:11 AM

This is also a problem with other modules. I have tried ESV and NKJV, both have similar issues. Others may too.

#9 Randy Steffens Jr

Randy Steffens Jr

    Silver

  • Active Members
  • PipPipPip
  • 173 posts
  • Accordance Version:10.x

Posted 31 May 2013 - 10:39 AM

Getting the same result in tagged and non tagged texts, and often the word "were" is not counted regardless of module...

Edited by rsteffens, 31 May 2013 - 10:40 AM.


#10 Helen Brown

Helen Brown

    Mithril

  • Admin
  • 7,988 posts
  • Gender:Female
  • Location:heart in Israel
  • Accordance Version:10.x

Posted 31 May 2013 - 04:08 PM

It turned out to be a bug, will be fixed when 10.1.7 is released, any day now.


Helen Brown
OakTree Software




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users