[resolved]Clause Search - Confused by Results - can anyone explain this?

Page 2 of 2 (38 items) < Previous 1 2
This post has 37 Replies | 4 Followers

Posts 804
Richard Lyall | Forum Activity | Replied: Tue, Oct 28 2014 8:04 PM

Hi Todd ... the thing is that Darrell was able to carry out an NT search on Greek NT SBL edition using a query which included "verb: to weep", if you have a look as his screengrab.

Posts 4595
Forum MVP
Fred Chapman | Forum Activity | Replied: Tue, Oct 28 2014 8:05 PM

I do see. based on Todd's post that you can run the sample search "subject:A Man verb:to bury" in the LHB and it works.

Posts 2066
LogosEmployee

Richard Lyall:

So what is the difference between "weep" and "weep; cry" as search terms\parameters, because that is not at all intuitive for an end user who doesn't know the software's inner workings!

When doing a clause search with a "verb:" you are actually specifying the original language lemma that is to be searched. The auto-completer provides you with the gloss of each lemma.

"weep" is the gloss of the lemma δακρύω
"weep; cry" is the gloss of the lemma κλαίω

You would see the same number of results if you instead did your searches as "person:jesus verb-lemma:κλαίω" and "person:jesus verb-lemma:δακρύω".

As was stated, unless you are looking for particular lemmas, you will probably have better luck finding what you are looking for using the sense: "person:jesus sense:to cry (tears)"

Posts 4595
Forum MVP
Fred Chapman | Forum Activity | Replied: Tue, Oct 28 2014 8:09 PM

Andrew Batishko (Faithlife):

When doing a clause search with a "verb:" you are actually specifying the original language lemma that is to be searched. The auto-completer provides you with the gloss of each lemma.

"weep" is the gloss of the lemma δακρύω
"weep; cry" is the gloss of the lemma κλαίω

You would see the same number of results if you instead did your searches as "person:jesus verb-lemma:κλαίω" and "person:jesus verb-lemma:δακρύω".

As was stated, unless you are looking for particular lemmas, you will probably have better luck finding what you are looking for using the sense: "person:jesus sense:to cry (tears)"

Thanks Andrew. Can you say why Darrell's search worked in 5.2 and it did not work in 6.0. Also the search Richard posted that returned Jesus looking in Acts is wierd

Posts 752
LogosEmployee
Sean Boisen | Forum Activity | Replied: Tue, Oct 28 2014 8:12 PM

Richard Lyall:

What I really want is some way to search for all weeping activity, however Logos wants to classify it, without having to guess the underlying semantic taxonomy first.

Probably the best way to do this is to search using the BSL senses with the new Sense search syntax. While there is a sense "to weep" (logos4:Senses;KeyId=ws.weep.v.02), the more general "to cry (tears)" (logos4:Senses;KeyId=ws.cry.v.02) probably captures your interest more accurately. (See https://community.logos.com/forums/t/92763.aspx for more on this data type search feature)

Using Bible Search in just the Gospels (logos4:Search;kind=Bible;q=$3CSense_to_cry_(tears)$3E;match=stem) returns 30 results in 22 verses, including John 11:35, Luke 19:41, and Luke 23:28.

Posts 1275
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Tue, Oct 28 2014 8:14 PM

Hi Richard.

Richard Lyall:

So what is the difference between "weep" and "weep; cry" as search terms\parameters

In the NT, these are different glosses for different underlying lemmas. As I understand it, the English gloss of the verb is the item that makes it into the 'verb:' operator. It is essentially running a search on the Greek through the cipher of the English gloss.

Instead, you probably want to rely on the 'sense:' operator to search using the senses of the Bible Sense Lexicon, which would be more reliable than English glosses. And tie in semantic roles while you're at it.

agent:Jesus sense:to cry(tears)

That returns two hits, Lk 19.41-44 and Jn 11.35. Note your Lk 23.28 instance is not included as Jesus is not the one weeping there.

Does that help?

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 804
Richard Lyall | Forum Activity | Replied: Tue, Oct 28 2014 8:16 PM

That makes sense. I wasn't clear on the connection between the Clause Search terms and the underlying lemmas.

And since each Sense can represent multiple lemmas, then it allows more intuitive searching where the query is expressed along the lines of "find all the places where Jesus wept".

Posts 2066
LogosEmployee

Richard Lyall:

Thanks Fred. In the meantime, more Clause Search weirdness.

Unless I'm mistaken, Jesus was definitely not the one doing the looking in Acts 27:15!!!

There is something odd going on there. This search works as expected: "sense:to face ⇔ to look in the eyes person:Jesus", but the order you provided does not. As I understand things, these should give the same results, so this looks like a bug. I'll send myself an email to remind me to write up a case for this when I get into work tomorrow.

Posts 804
Richard Lyall | Forum Activity | Replied: Tue, Oct 28 2014 8:22 PM

That's making more sense now - thanks for your input.

However, I've run into another confusing thing when carrying out the following:

  1. Right click on "wept" on Jn 11:35; ESV
  2. Select "To cry (tears)" from the flyout
  3. Select "Search Inline\Search this resource" on the left of the flyout

This launches a Bible search with <Sense = to cry (tears)> query, which only returns Jn 11:35, whereas running a Clause Search with sense:to cry (tears) returns 40 results.

Is this expected in some way - am I missing something, or is this a bug?

Thanks
Richard

Posts 2066
LogosEmployee

Richard Lyall:

That's making more sense now - thanks for your input.

However, I've run into another confusing thing when carrying out the following:

  1. Right click on "wept" on Jn 11:35; ESV
  2. Select "To cry (tears)" from the flyout
  3. Select "Search Inline\Search this resource" on the left of the flyout

This launches a Bible search with <Sense = to cry (tears)> query, which only returns Jn 11:35, whereas running a Clause Search with sense:to cry (tears) returns 40 results.

Is this expected in some way - am I missing something, or is this a bug?

Try changing the Bible search to "<Sense to cry (tears)>". This gives 168 results for the entire Bible, or 47 results in just the New Testament. The = operator causes the search to only match that exact sense, and not any of the child senses (use the Bible Sense Lexicon for details on child senses).

I don't have an explanation yet on why this gives 47 results in the NT instead of 40 like clause search. The added results appear to be correct, so it seems clause search is missing some results, and I'm not sure why. It could be a bug, but I won't be able to investigate tonight.

Posts 1275
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Tue, Oct 28 2014 9:56 PM

Hi Richard.

Richard Lyall:

Thanks Fred. In the meantime, more Clause Search weirdness.

Unless I'm mistaken, Jesus was definitely not the one doing the looking in Acts 27:15!!!

So, what you're really searching for is where there is a reference to the person 'Jesus' in a clause that also uses the sense "to face <=> to look in the eyes." If you wanted to specify Jesus doing the looking, you'd use semantic roles: agent:Jesus, not person:Jesus. If you want to find "what did [participant] do" sorts of things, then the agent semantic role is usually what you're looking for.

That said — I have no idea why the reference to 'Jesus' is happening here, or why it shows up. Note that in that front position, I can put just about anything (including "agent:Jesus") and a result to this verse will fire, so this sounds more like a bug processing the query (perhaps due to the '<=>' character, aka the 'spaceship' operator for you Perl hackers?)  I'll file a report on that.

When I do a BWS on ἀντοφθαλμέω, there are no clause participants, which tells me again that something weird is happening in query processing. 

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 1275
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Tue, Oct 28 2014 9:58 PM

Rick Brannan (Faithlife):
I'll file a report on that.

Since Andrew said he'd look into it, I'll let him file the report.

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 1275
LogosEmployee
Rick Brannan (Faithlife) | Forum Activity | Replied: Tue, Oct 28 2014 10:15 PM

Andrew Batishko (Faithlife):
I don't have an explanation yet on why this gives 47 results in the NT instead of 40 like clause search. The added results appear to be correct, so it seems clause search is missing some results, and I'm not sure why. It could be a bug, but I won't be able to investigate tonight.

I think it is either a difference in the way that 'results' are counted in each search (multiple hits in same clause/same verse perhaps counted differently) or it may have to do with how clauses and overlapping data is represented in the clause search database.

I sent the ESV results to a passage list, then sent the SBLGNT results to a different passage list. Then I used the cool Logos passage list set math (in since L5 -- haven't used it? you should, try 'Merge' and select a different list) to find the "Symmetric difference" between the sets. The symmetric diff involves:

  • Mk 5:39 (ESV, not SBLGNT)
  • Mk 16.10 (ESV, not SBLGNT)
  • Lk 19:41 (ESV, not SBLGNT)
  • Lk 19:41-44 (SBLGNT, not ESV).

My guess is that the differences have to do with how the hierarchical nature of the clause structure is dealt within in the encoding of data for the cause search. I'll file a case to look into it further.

Rick Brannan
Data Wrangler, Faithlife
My books in print

Posts 26263
Forum MVP
MJ. Smith | Forum Activity | Replied: Tue, Oct 28 2014 10:34 PM

Of course I use the passage list set functions ... and I know Dave does as well ... for exactly the sort of thing you have.Stick out tongueWink

Orthodox Bishop Hilarion Alfeyev: "To be a theologian means to have experience of a personal encounter with God through prayer and worship."

Posts 2066
LogosEmployee

Andrew Batishko (Faithlife):

Try changing the Bible search to "<Sense to cry (tears)>". This gives 168 results for the entire Bible, or 47 results in just the New Testament. The = operator causes the search to only match that exact sense, and not any of the child senses (use the Bible Sense Lexicon for details on child senses).

I don't have an explanation yet on why this gives 47 results in the NT instead of 40 like clause search. The added results appear to be correct, so it seems clause search is missing some results, and I'm not sure why. It could be a bug, but I won't be able to investigate tonight.

Oops, do the Bible search on the SBLGNT resource. This will give back 41 results not the 47 I was getting from the ESV search I was doing earlier. I've created a case to investigate the difference of one result from doing a clause search.

Posts 49
LogosEmployee
Chris Robbers | Forum Activity | Replied: Wed, Nov 5 2014 4:20 PM

This has been fixed and will be included in an upcoming release. It only affected sense terms on the right side of a compound query.

Posts 804
Richard Lyall | Forum Activity | Replied: Wed, Nov 5 2014 4:39 PM

Great thanks.

Posts 1471
Angela Murashov | Forum Activity | Replied: Tue, Nov 18 2014 2:55 PM

Chris Robbers (Faithlife):
This has been fixed and will be included in an upcoming release.

This should be fixed in 6.0a RC1. 

Page 2 of 2 (38 items) < Previous 1 2 | RSS