Redesign Syntax Search?

I will confess: I find syntax searching intimidating. I have had a few attempts with it, with some degree of success. But part of the problem, at least for me, is that the interface and the way the parameters work are too different from the other types of searches (notably clause and morph).

I am wondering if there would be any way (and will) at all to redesign the concept so that users will find themselves in a more recognisable environment. For instance, could something similar to the morphology picker be used to build the search?

Syntax search experts can give their opinion as to whether this could work, whether it would be a good thing or not, etc.

I'd be curious to hear also from many users whether they actually use syntax search or -- like me -- might want to but find it too difficult to use.

Find more posts tagged with

Comments

Graham Criddle

I'd be curious to hear also from many users whether they actually use syntax search or -- like me -- might want to but find it too difficult to use.

I have used it quite a bit in the past but not much recently

And I was interested by Devin Roza's comment (https://community.logos.com/forums/p/130477/866874.aspx#866874) where he finds less and less need to use it. Are there specific things you want to do with it that you can't do using other Logos tools?

Francis

https://community.logos.com/discussion/comment/868796#Comment_868796

Again, as someone who actually does not use it a whole lot, I can only say that it seems to me that while we can modulate other searches to look for constructions, word order, etc., I don't think we can incorporate arguments in these that take into account syntactical force.

Dave Hooton

https://community.logos.com/discussion/comment/868802#Comment_868802

I don't think we can incorporate arguments in these that take into account syntactical force.

The Greek/Hebrew Grammatical Constructions datasets offer an easier ride for some cases e.g. {Section <GramCon GenAbs>}, but Syntax Search has always suffered from controlling results (e.g. it shows a result in multiple overlapping clauses) and the uncertainty of completeness (e.g. the "gaps" phenomenon, allowing for a different order of elements, or exclusion of some intervening elements). There could be a higher level of specification but my requests have not been considered.

Clint Cozier

The other problem (or so it seems to me) is that the additional texts needed to make syntax searching really useful are not being tagged. At a minimum, I'd want a tagged LXX and a scattering of Philo and or Josephus before it would be really useful. Even when I use syntax searches in the NT, I still end up translating the search into a morphological search to work in the LXX, church fathers (and mothers) and semetic greek. That I know of, no one is working on expanding the texts that have syntax tagging beyond Oak Tree finishing up their projects. Fh or a while, FL had a pre-pub to do the LXX with the Cascadia syntax schema, but that fell by the wayside. We have quite a nice body of morph tagged texts, I'm happy to see Logos working on that interface and building something very exciting that takes advantage of what we have now.

I'm quite impressed with the interface they are building with the Morph Query funcationality in Logos Now. My hope is that in that single interface they can wrap both morph and syntax searching into a single tool much like the construct search interface in Accordance.

Francis

https://community.logos.com/discussion/comment/868863#Comment_868863

The Greek/Hebrew Grammatical Constructions datasets offer an easier ride for some cases e.g. {Section <GramCon GenAbs>}

It would be interesting to see whether Grammatical Constructions can be sufficiently expanded to make Syntax search redundant and the latter be phased out.

My hope is that in that single interface they can wrap both morph and syntax searching into a single tool much like the construct search interface in Accordance.

I don't know what the Accordance model looks like, but I assume staff at Faithlife look at what competition is doing and can get some ideas from that. Since Syntax search works with clauses as well, it would make sense if there was also some kind of conjunction with Clause search.

Fr Devin Roza

https://community.logos.com/discussion/comment/868952#Comment_868952

Faithlife has done a LOT already to minimize the need for Syntax search. A lot of the new datasets are for syntax tagging, which is great, as search extensions are much easier to use than Syntax Search. And, as Clinton mentioned, the Morph Query Builder goes a long way as well.

Apart from continuing to add new datasets of typical Syntax Search case uses, the main thing I'd like to see Faithlife do to further reduce the need for using Syntax Search is:

Allow searching across multiple verses on the Bible search tab.
Allow searching within Clause boundaries on the Bible search tab.
Integrate the searches which are limited to the Clause tab into the Bible search tab using search extensions. Ideally, the clause search tab should simply be eliminated, once clause boundary is allowed for on the Bible search tab.
Improve the quality of the current Clause search tagging, especially in the data taken from the Andersen and Forbes data. It doesn't seem to me to be trustworthy enough. Here's an example of what I'm talking about... my experience has been that the OT "subject" tagging is particularly in need of improvement.

If I understand correctly, doing the changes in 1-3 would imply major rewrites of both code and some tagging. Hopefully even so it can happen someday. Regarding number 4, I wonder if Faithlife did a little review to get together some example of bad "subject" and similar, which I'm guessing is a result of some automatic tagging done by means of scripting, and presented it to Andersen and Forbes, if they might be willing to go through and correct the errors. Other problems are from the gap problems in the Syntax Searchd data, but maybe a manual review of those could be done, by limiting manual review to verses with gaps.

Francis

https://community.logos.com/discussion/comment/868982#Comment_868982

Allow searching across multiple verses on the Bible search tab.

Allow searching within Clause boundaries on the Bible search tab.

How would you envision it? How would we pick multiple verses or clause boundaries in the UI?

Integrate the searches which are limited to the Clause tab into the Bible search tab using search extensions. Ideally, the clause search tab should simply be eliminated, once clause boundary is allowed for on the Bible search tab.

I like the idea of using search extensions and all the more as tabs isolate a type of search from other types limiting the ability to combine them. But there is also a downside to it: the tabs make it simpler for many users to keep the type searches apart. Combining tabs might put more emphasis yet on learning to write the proper search parameters. Likewise multiplying datasets and search extensions increase the expertise needed to use searches effectively and increases the potential for user mistakes. Each tabs has also its own quick helps which could not all be there if combined. One possible solution would be to move more toward pickers (like morph search) as opposed to writing parameters, so long that one is not too constrained by the pickers to take full advantage of the combination possibilities. Better yet: have both. Have pickers as the "easy track" and search syntax for the experts. There is software out there that offers alternate UI based on level of expertise.

Rick Brannan (Logos)

https://community.logos.com/discussion/comment/868863#Comment_868863

FL had a pre-pub to do the LXX with the Cascadia syntax schema, but that fell by the wayside.

In the 6.x Logos Now release cycle (6.12?) we did release a syntactic analysis of the deuterocanonical/apocryphal portions of the LXX.

https://www.logos.com/product/64904/cascadia-syntax-graphs-of-the-lxx-deuterocanon-apocrypha

I do not anticipate that we will analyze the balance of the LXX for any release in the near or mid-term.

Fr Devin Roza

https://community.logos.com/discussion/comment/868994#Comment_868994

Allow searching across multiple verses on the Bible search tab.

Allow searching within Clause boundaries on the Bible search tab.

How would you envision it? How would we pick multiple verses or clause boundaries in the UI?

I envision having an option next to the Bible resource dropdown menu (or anywhere on that line) that would specify range, with options like "verse", "clause", or "chapter." We can already search by chapter on the Basic tab, and by clause on the Clause tab, and by verse on the Bible tab. Ideally on the Bible tab we could choose which boundary we prefer.

One possible solution would be to move more toward pickers (like morph search) as opposed to writing parameters

Yes, I agree. I think a combination of a GUI dropdown picker to choose search extensions or datatypes, as well as continued improvements to autocomplete to suggest search extensions, datatypes, to automatically suggest what is available based on where you are in the search string as you move around (as is common in GUIs for programming languages).

I actually hope the Bible Browser included in Verbum and Logos Now is a first step towards a GUI for search. You could imagine a sort of "magic wand" on the search pane, and when click on it, something like the left hand pane of the Bible Browser tool drops down. You select the search extension, datatype, etc. you want and it would paste it wherever your cursor was in the search string.

Here is a picture of the left hand pane of the Bible Browser so you can see what I mean. In this case, if I were to click on "Declarative", it would paste the text {Section <Sentence = Declarative>}. Of course, this dropdown would not be identical if it were on the search panel, but the basic idea is the same, and much of the data and options would be as well!

PetahChristian

https://community.logos.com/discussion/comment/869351#Comment_869351

I envision ...

I actually hope ...

[Y] Yes, please, to everything.

Typing out obscure search syntax is off-putting, but I truly enjoy learning and discovering unexpected results via an interactive UI.

Francis

https://community.logos.com/discussion/comment/869355#Comment_869355

One would think that if we move toward a browser type of solution, it would take care of text boundaries as well since it would have to be one of the categories one can pick from. There would need to be some enlargement so that it applies to more than Bible resource types so as to include any other type of material that is tagged with datatypes and indices. So perhaps Basic and Bible could be combined in one browser?

Of course, it would be even better if all the searches were accessible through one UI, that is, to include clause and morph parameters picking as well. The question is whether it would be possible to do all that without overcrowding the interface (not to mention that in time, other datasets may be developed and added). Finding a way to work in all the possible arguments and give the user flexibility to combine them was part of the difficulty in Syntax search (although imo the browser model is more intuitive than working with nodes).

I agree that improving autocomplete would help. I find that it does rather well with single terms (e.g., Abraham will suggest person, topic, etc) but not as well when one attempts to build a complex search.

I have suggested "Tell me what you want to do" functionality before (Microsoft Office style), and am surprised that it seems to have been met with little interest, although I think it could help a great deal as well.