BUG: Rcik, Bradley, Sean re: clause search et. al.

Take:

The values searched here are, as I understand it, the "lexical value" fields from the Lexham Greek NT Interlinear.

Those values are context sensitive; when the editor determined more
than one sense might apply, he listed them all. It is not a simple
concordant gloss.

It seems the expected behavior for clause search is more of a 1:1
correspondence between the lemma and the gloss. I believe (Sean can
clarify as he sees fit) that we're discussing ways to do this
internally.

I think one thing to realize through all of this is that clause
search is really cool, but nothing like it has really ever been done
before. We're learning, just like you, how to fine-tune things. Feedback
like this helps in that process, so it is greatly appreciated.

Now look at:

What Rick is saying the users think it the reverse of what is happening.

1. the example on the left has two lemmas and one sense.

2. the example on the right has one lemmas and two senses.

Logos has a big problem - it is treating the right hand side as a single sense.

This is not what we want:

It seems the expected behavior for clause search is more of a 1:1
correspondence between the lemma and the gloss.

What we want is the 1:many treated as 1:many.

============================

Would one of you also help sort out the source of each section of a BWS - source in the sense of:

lemma with homograph indicator
lemma for all homographs
roots of lemma with homograph indicator
roots for all homographs
differences, if any, between Greek & Hebrew, nouns & verbs ...

Every time I think I've got it sorted out, I try another word that blows that theory out of contention.

Find more posts tagged with

Comments

Dave Hooton

Rick's quote comes from http://community.logos.com/forums/p/60050/427026.aspx#427026

Also see http://community.logos.com/forums/t/59899.aspx

Jon

1. the example on the left has two lemmas and one sense.

It shouldn't though right, since ειπον is the (second) aorist form of λεγω?

It appears that there a bunch of inconsistencies here with how different verbs that take different forms/roots for different principle parts have their lexical forms tagged. In MJ's example, aorist forms of λεγω (and future, perfect, plu-perfect) are tagged with εἰπον as their lemma. Similarly for ὁραω, the aorists are tagged as εἰδον (the future forms from ὀψομαι are tagged as ὁραω); however for ερχομαι all of the tense-forms from different principle parts are tagged as ερχομαι...

Note: There is one instance of an aorist of λεγω being tagged as λεγω rather than ειπον — the SBL textual variant in John 21:17.

[:S]

Bradley Grainger (Logos)

Would one of you also help sort out the source of each section of a BWS - source in the sense of:

I'm not sure what you're asking for; can you please clarify? (And it's not clear to me what the phrase "lemma for all homographs" would even mean.)

Every word in our morph-tagged Bibles is tagged (in context) with one or more lemmas. These lemmas come from various morphological analyses provided by various sources; for "Logos" morph, it's an in-house analysis. Lemmas that would otherwise be homographs have a number appended to make them unique. Each lemma is, in the abstract, associated with one or more roots.

This many-to-many mapping of word <-> lemmas <-> roots is grouped and summarised in various ways in the different sections. A BWS (in Greek or Hebrew) starts with a single lemma, so the Translations sections show the lemma -> words associations, the Roots section shows the lemma <-> roots associations, etc. A lemma is the key that ties the whole guide together.

Dave Hooton

https://community.logos.com/discussion/comment/427782#Comment_427782

Would one of you also help sort out the source of each section of a BWS - source in the sense of:

I'm not sure what you're asking for; can you please clarify?

As background, see

http://community.logos.com/forums/t/59574.aspx

MJ. Smith

https://community.logos.com/discussion/comment/427782#Comment_427782

A lemma is the key that ties the whole guide together.

First, there are threads in which users have apparently entered a lemma without a homograph number to distinguish it.

Second, the results in some cases do not appear to be correctly limited to the homograph entered. Dave has directed you to one (of several) threads that illustrates this.

Third, I apparently misunderstood your answer to a question re: the roots. I'll play with some of the problematic cases again.

In the clause search is where none of us have been able to sort out the one-to-many relationship. The Search seems to treat it as a one-to-one. Could you give us a word to test that has a one-to-many so that we can prove to ourselves that the clause search is actually working.

I assume that I have found a lemma assigned 3 senses here:

but the clause search treats "to speak, to say, to tell" as a single sense. If it is a single sense then I need to understand how it differs from "to say, to tell" which is also treated as a single sense.

In the mode of full disclosure, I'm used to Sanskrit in which the dictionary entry is the root and the concept of word/lemma tends to fall apart under the extreme use of compounding. Hence I think in terms of root -->stem-->(compound)-->grammatical affixes --> surface text

Jon

https://community.logos.com/discussion/comment/427967#Comment_427967

This certainly is confusing: [:)]

Is this right?

The options when searching for an English verb in clause search come from the "English Lexical Value" in the interlinear.
These seem to function like "senses" though there are no verbs in the sense lexicon yet. So if you pick the right English "sense" it will map to multiple lemmas (e.g. "to say").
λαλεω maps to "to speak, to say, to tell", "to speak", "to speak, say", "to say, tell", "to talk, speak", "to say"
λεγω/ειπον maps to "to say", "to say, speak"
φημι maps to "to say", "to declare, say"

If anyone can work out how the different "senses"/English Lexical Values assigned to λαλεω have different meanings you're doing better than I [:)] ...it feels like you needed the sense lexicon before you did the tagging with English Lexical Values if they were to be tagged consistently.

How I'd expect this to work based on the way the feature is advertised is that it should be useful for someone with no Greek or Hebrew — if I wanted to find all the times Jesus "said" something that should find all relevant occurrences without needing to search for 10 variations on say/speak/tell etc.

Bradley Grainger (Logos)

https://community.logos.com/discussion/comment/427967#Comment_427967

I assume that I have found a lemma assigned 3 senses here:

...

but the clause search treats "to speak, to say, to tell" as a single sense. If it is a single sense then I need to understand how it differs from "to say, to tell" which is also treated as a single sense.

The "verb" field in Clause Search simply searches the contents of the "English lexical value" line in LGNTI. It has nothing to do with "senses" or BSL or the lemma or anything else. It's essentially the same as doing a Basic/Bible Search in "Lexical Value" in LGNTI for the phrase "to speak to say to tell".

If you do this, you'll only get 26 results, not 27. I've reported the doubled hit (in Clause Search) in Acts 3:22 as a possible bug.

Bradley Grainger (Logos)

https://community.logos.com/discussion/comment/427967#Comment_427967

but the clause search treats "to speak, to say, to tell" as a single sense. If it is a single sense then I need to understand how it differs from "to say, to tell" which is also treated as a single sense.

I've read some more of your posts on other threads (I'm quite behind on the forums), so I think I may be starting to understand the confusion.

The Clause Search engine essentially performs an exact text match using the text that follows "verb:", comparing that text against the full contents of the cell for the "English lexical value" line in the LGNTI Interlinear. The comma is not a search operator; it's part of the exact text match.

So "to say" doesn't match "to say, to tell" or "to speak, to say, to tell", because it's looking for exact string matches, not substrings.

So a correction to my last post is necessary; it's not quite the same as searching "Lexical Value" in Basic Search, because that will find "to speak, to say, to tell" when you just search for the phrase "to say".

Bradley Grainger (Logos)

https://community.logos.com/discussion/comment/428822#Comment_428822

The Clause Search engine essentially performs an exact text match using the text that follows "verb:", comparing that text against the full contents of the cell for the "English lexical value" line in the LGNTI Interlinear. The comma is not a search operator; it's part of the exact text match.

This will be changing in 5.0b (but I haven't seen or run the new code yet, so I don't have further details).

For now, it's probably best to treat this as a historical definition of how Clause Search worked in 5.0 and 5.0a.

Dave Hooton

https://community.logos.com/discussion/comment/428833#Comment_428833

This will be changing in 5.0b (but I haven't seen or run the new code yet, so I don't have further details).

Thank you for that good news.

Tim Hensler

https://community.logos.com/discussion/comment/428833#Comment_428833

The Clause Search engine essentially performs an exact text match using the text that follows "verb:", comparing that text against the full contents of the cell for the "English lexical value" line in the LGNTI Interlinear. The comma is not a search operator; it's part of the exact text match.

This will be changing in 5.0b (but I haven't seen or run the new code yet, so I don't have further details).

For now, it's probably best to treat this as a historical definition of how Clause Search worked in 5.0 and 5.0a.

Thank you for your help, Bradley, but I think you should go home now. You're starting to answer your own posts. Usually not a good sign. [:S]

Bradley Grainger (Logos)

https://community.logos.com/discussion/comment/428873#Comment_428873

You're starting to answer your own posts. Usually not a good sign.

Just in case it wasn't clear, I was correcting a statement I made in an earlier post, and quoting it for context. (Some people subscribe to the forum via email, and edits to existing posts don't get sent out, only new posts. Thus, I usually only make minor edits to existing posts and create a new post for a major edit.)