ATTN: Bradley: Can you help me understand why NOT INTERSECTS doesn't work in one case?

While helping another user, I discovered these two searches gave very different results:

(<LogosMorphGr ~ V??P????> NOT INTERSECTS <Root = lbs/el/λεγω>) BEFORE 1 WORD <LogosMorphGr ~ V??M????>

and

(<LogosMorphGr ~ V??P????> NOTEQUALS <Root = lbs/el/λεγω>) BEFORE 1 WORD <LogosMorphGr ~ V??M????>

With the latter being correct.

The results from NOT INTERSECTS are so odd, it's almost as if Logos as misinterpreted the syntax. Is that a bug, am I just going beyond the design of the parser, or have I misunderstood something?

Find more posts tagged with

Comments

Andrew Batishko

INTERSECTS works by finding results for the left term and the right term, and then returns verses that contain both terms where those terms overlap.

NOT INTERSECTS works by finding results for the left term and the right term, and then returns verses that contain both terms where those terms do not overlap.

EQUALS works by returning the verses where a word matches the left and the right term. This combines the two terms into a single term.

NOTEQUALS works by returning the verses where a word matches the left but not the right term. This combines the two terms into a single term.

Because the EQUALS and NOTEQUALS operators combine their terms into a single term, you only get two different colors of result highlighting in your NOTEQUALS search compared to three with the NOT INTERSECTS search.

Mark Barnes

https://community.logos.com/discussion/comment/973489#Comment_973489

Thank you Andrew.

That doesn't explain why I get 24 verses returned with NOT INTERSECTS, but 59 verses with NOTEQUALS. And most of those 24 verses are wrong.

For example, Matthew 15:25 is returned with NOT INTERSECTS. The highlighted words are ἐλθοῦσα (which matches the first term, as it's a participle), βοήθει (which matches the last term, as it's imperative, although it's much more than one word away). Also highlighted is προσεκύνει, which is indicative and doesn't match any term!

Andrew Batishko

https://community.logos.com/discussion/comment/973523#Comment_973523

The problem isn't the INTERSECTS operator. The problem is that the BEFORE operator only supports certain types of complex (containing an operator) terms on its left and right sides.

This will work: (a OR b) BEFORE 1 WORD c

This will not: (a AND b) BEFORE 1 WORD c

INTERSECTS and NOT INTERSECTS have the same problem.

Because the syntax isn't supported, the actually query is the same as: (<LogosMorphGr ~ V??P????> NOT INTERSECTS <Root = lbs/el/λεγω>) AND "BEFORE" AND <LogosMorphGr ~ V??M????>

We are aware that it would be nice if the UI would show you when the query is not valid, but I don't know when or if such a feature will be implemented.

Keep Smiling 4 Jesus :)

I discovered these two searches gave very different results:

(<LogosMorphGr ~ V??P????> NOT INTERSECTS <Root = lbs/el/λεγω>) BEFORE 1 WORD <LogosMorphGr ~ V??M????>

Likewise encountered this surprise search transformation into

Because the syntax isn't supported, the actually query is the same as: (<LogosMorphGr ~ V??P????> NOT INTERSECTS <Root = lbs/el/λεγω>) AND "BEFORE" AND <LogosMorphGr ~ V??M????>

Expected search evaluation was parenthetical evaluation first followed by result set passed to BEFORE for further filtering.

Keep Smiling [:)]

Mark Barnes

https://community.logos.com/discussion/comment/973553#Comment_973553

This will work: (a OR b) BEFORE 1 WORD c

This will not: (a AND b) BEFORE 1 WORD c

INTERSECTS and NOT INTERSECTS have the same problem.

I don't understand this. At least, I do understand why (a AND b) doesn't work — because (a AND b) doesn't have a specific location that could be before something else. But I don't understand why NOT INTERSECTS doesn't work in that context. (a NOT INTERSECTS b) does have a specific location — the location of a.

I guess what I'm saying is "Because NOT INTERSECTS will only return one term, why doesn't it function the same way as NOTEQUALS?". For that matter, why do you say that NOTEQUALS combine the two terms into one, when only one term matches anyway?

In addition, this comment seems to be incorrect:

NOT INTERSECTS works by finding results for the left term and the right term, and then returns verses that contain both terms where those terms do not overlap.

It says that NOT INTERSECTS requires BOTH terms to be present but not overlapping. That's incorrect. My understanding is that NOT INTERSECTS requires EITHER both terms to be present but not overlapping, OR the first term to be present without the second.

I'm sorry for asking these detailed questions, but given that the UI or documentation doesn't help, it's important to have 100% clarity about this functionality.

Todd Phillips

https://community.logos.com/discussion/comment/973610#Comment_973610

This will work: (a OR b) BEFORE 1 WORD c

This will not: (a AND b) BEFORE 1 WORD c

INTERSECTS and NOT INTERSECTS have the same problem.

I don't understand this. At least, I do understand why (a AND b) doesn't work — because (a AND b) doesn't have a specific location that could be before something else. But I don't understand why NOT INTERSECTS doesn't work in that context. (a NOT INTERSECTS b) does have a specific location — the location of a.

I guess what I'm saying is "Because NOT INTERSECTS will only return one term, why doesn't it function the same way as NOTEQUALS?". For that matter, why do you say that NOTEQUALS combine the two terms into one, when only one term matches anyway?

In addition, this comment seems to be incorrect:

NOT INTERSECTS works by finding results for the left term and the right term, and then returns verses that contain both terms where those terms do not overlap.

It says that NOT INTERSECTS requires BOTH terms to be present but not overlapping. That's incorrect. My understanding is that NOT INTERSECTS requires EITHER both terms to be present but not overlapping, OR the first term to be present without the second..

The last part I underlined is how I would expect and want it to work.