Accordance Analytics: Questions, Issues, Reminders, Requests

[mgvh]

I have been working with Accordance Analytics, and I have identified some questions, issues, reminders, and requests.

As an example, do a search in the NA28 for ἰδού.

• Open Hits Graph (default display is Hits per 1000 words; I'm starting with bar chart)
  • What does "Hits per 1000 words" mean in a book like Jude with less than 1000 words?
• At first glance, it looks like same density / frequency of hits in Jude as in Galatians, but there is only one hit in each even though Galatians is ~5x longer
• Looking more closely, it looks like there is a split line with Galatians > click hold on bar to get the reference, and it looks like Galatians 1.20 is counted twice
  • > I use the zoom in icon, and now I see that there are 3 bars
    
  • But if I click hold to get references, it only shows two hits: 2 Cor 12.14 and Gal 1.20 >> How is Accordance calculating “1000 words”?
    It look like it is just counting every 1000 words beginning with Matthew 1.1 and ignoring book boundaries. True?? But why 3 bars?

    o    I get similar looking results without zooming by spreading out graph across two monitors, and again I see 3 bars marking Galatians 1.20

  • If I switch to Line view, I see that there are 3 dots for Galatians 1.20

• So, I reduce “Hits per” down to per 100 words, and now there is only one mark for Galatians 1.20, but it still appears to be same density / frequency as Jude
  • >> Apparently, any frequency under 1.0 / X words is graphed as 1
  • I.e., the graph does not provide enough detail to give a clear view.
• >> I try using Table Analytics, and now I can see that “Hits per 1000 words” in Galatians is .45 (1x in a book with 2232 words); in Jude 2.17 (1x in book with 460 words).
  • >> This is technically correct in saying it is measuring “Hits per 1000 words,” but what it really is calculating “Hits per total words in book” (That would be a clearer descriptor for all the NT books with less than 100 words: 2Thess, Titus, Philemon, 2John, 3John, Jude) >> Great! That is what I really want.
  • I can also choose to “Show Chapter Detail” which gives a more precise look
• >> Now I open Table Bar Chart, and it does show relative frequency correctly, BUT…
  • Chart can only be displayed with horizontal bars
  • There is no way to display books with no hits, so I don't really have an overall clear view of distribution in the NT
• >> So I tried using the results from the Table and creating my own chart by copy/pasting into Excel, but that’s a bunch of work, and it still doesn’t specifically list books with no hit results, but this is the kind of chart I was trying to generate
  • 

SUMMARY:

Reminders

• The Hits Graph is a helpful but blunt tool. It is trying to show density/frequency (hits per X words), but it can be misleading when using a large X (e.g. 1000), and it is not precise enough when using a small X (e.g. 100) because everything under “1 per 100” graphs the same as “1.” It is most helpful in showing large trends across a book and across genres (Gospels, Acts, Pauline…).
• The Table Bar Chart is what should be used for accurate frequency / density results “per book” and “per chapter”

Suggestions / Requests

• For the Hits Graph, it would be good to have the vertical axis give more precision than rounding to whole numbers
• For Table and Table Bar Chart: Change label of “Hits per 1000 words” to “Hits per total words in book” (and “Hits per total words in chapter”)
• Provide option to “Display books with no hits” in Table and Table Bar Chart analytics
• Provide option to have vertical bars with the Table Bar Chart

 

Note that my observations here partly arise out of my previous use of BibleWorks. It was never known for attractive graphing functions, but it had more options for graphing, e.g., detail could be book, chapter, or verse; zero hit results could be displayed or no; axis could be switched to horizontal/vertical; hit could be calculated by percent or hit and by number of words in book, verses, or version

 

  

[Donald Cobb]

This thread has been out there for a while, but I'm only seeing it now. Thanks Mark. I too would like to see this tool refined a little more. For me it's too approximate to be really useful other than to give a "quick and dirty" idea of vocabulary usage. I've already commented on this here and here. To date, the most useful part of this tool is the number of hits, rather than the percentage of hits per xxx words. Can we hope for an improvement in Acc 14? 🙂