Bug: Search for Louw-Nida

Dave Hooton

The following search produces spurious results at Rev 8:13, 14.6 & 19:17

It should produce the same results as <ln1.10-12> i.e. 217/205. Note that ln 1.10 occurs 3 times

A worse doubling up of results occurs with <ln 1.11> OR <ln1.10-12>

Find more posts tagged with

Comments

Tonya J Ross

I'll report the issue to Development for investigation.

Bradley Grainger (Logos)

The counts are reported from a "search term" perspective. There are 217 results for the first term, and three for the second, for a total of 220 results. The hit highlighting shows you (in blue) that there are just three results for your second term.

Unifying the counts for "duplicate" results is on our tasklist for consideration, but there are a number of cases where this is not easy to implement (or even easy to define). If two different Greek words (combined with OR) both have a result as part of an English idiom in a reverse interlinear, is that one result, or two? What if the search results are also shown in parallel in a different resource that doesn't combine them into an idiom? Does it show both results, or just the merged one? Are they only merged if the results are precisely the same length? (And many other similar questions that would need to be asked and answered before we could start implementing.)

In this case, the search feels contrived, in that you're searching for "X OR Y" where Y is a subset of X. A similar textual search (which produces similar results) is: Jesus OR "Jesus Christ".

Dave Hooton

In this case, the search feels contrived, in that you're searching for "X OR Y" where Y is a subset of X. A similar textual search (which produces similar results) is: Jesus OR "Jesus Christ".

There are a few problematic searches but I thought I'd ask about a more obvious case! It's easily avoided, of course.

But here's an old one - why are results doubled in some Greek resources e.g. <lemma = lbs/el/ἀγαπάω> ANDEQUALS <lbs-morph+el ~ V???2???> is double in Elzevir 1624, Scrivener 1894 but not in Scrivener 1881 or Byz? Why is it double in UBS4 Int but not in UBS4 (both NA27's show double)?

Bradley Grainger (Logos)

But here's an old one - why are results doubled in some Greek resources e.g. <lemma = lbs/el/ἀγαπάω> ANDEQUALS <lbs-morph+el ~ V???2???> is double in Elzevir 1624, Scrivener 1894 but not in Scrivener 1881 or Byz? Why is it double in UBS4 Int but not in UBS4 (both NA27's show double)?

TR1881 and BYZ aren't morph resources, so I get 0 results for that search in them.

I'm not sure why the result counts aren't doubled in UBS4. AFAIK, all ANDEQUALS searches return the sum of the hits for both terms; I don't know why this wouldn't be the case in UBS4.

Dave Hooton

TR1881 and BYZ aren't morph resources, so I get 0 results for that search in them.

I get 36 in each

Scrivener 1881

LLS:SCRMORPH
2012-10-31T20:46:54Z
SCRMORPH.logos4

Byz 2005

LLS:1.0.331
2012-04-18T18:29:14Z
BYZPRSD.logos4