8 replies to this topic

[sljones]

Hello. I was at Accordance training today with J.P.Kang and it was great. I saw lots of new ways to use accordance I hadn't seen (or thought of) before.

BUT.

There's still one feature that you don't have whose absence baffles me, which is a "search within the results on this page" feature.

For example, if I do a search on epi in Thayer, I get a screen or two of results. Some of the results have a semantic significance (e.g., the scripture refs) but many do not, like "Alexander Buttmann" (whoever he is/was) because they're just plain text. You have a feature to add fields (with tools at least) so you can search for the former kind of tagged data, but not for random text. (Suppose Mr. Buttmann was listed in three places in this entry and I wanted to skip from one to the next. This is particularly relevant with the longer entries in some of the tools, or when there are many results.) (Another example: I do a search in the Greek text but I have a parallel NRSV pane that I want to skip through without doing some funky NRSV-search linked to the results of an existing NA27 search. I know--well, I assume--I can do that, but I want something super-interactive.)

Essentially, what you have now is a great search engine for finding things, like Google. What I would like to supplement it is a "help my eyes see what is in front of me" feature like Edit+Find in a web browser. In fact, I would recommend a user interface just like Firefox, where I hit Cmd+F or / and get a little mini-search window at the bottom of the pane. I type a letter and an incremental search takes place as I continue to type. Then when I'm done I can dismiss that sub-pane or keep it.

Here is an example:

In the example, the page from the Accordance website is analogous to the results of a search using all the cool Accordance features. Now I just want to find things in it other than my search parameters.

Edited by sljones, 25 October 2005 - 04:40 PM.

[David Lang]

I'm afraid I'm a little confused. You seem to be aware that you can search multiple fields at the same time, but you make the statement that you can search "tagged data" like Scripture references, but not plain text. That's simply not the case.

In Accordance tools, everything has been divided into a specific field, including "plain text." Typically, the body text of an article is placed into a "Content" or "English Content" field. To do your search for every occurrence of "Buttmann" in Thayer's article on "Epi," search the Greek Entry field for "Epi," then open the More Options section, add one additional field, choose "English Content" in the new Field pop-up menu, type "Buttmann," then click OK.

Every occurrence of the word "Buttmann" in that article will then be highlighted. To go from occurrence to occurrence, simply use the Mark buttons.

I know that's not exactly what you're asking for, but it should help you to find what you're looking for quickly.

[sljones]

Yes. I understand that Tools have this "add field" feature. The feature works as I sketched and you described in your response.

Texts do not have an "add field" feature, however, and one of the uses I had in mind is the other one I mentioned, where I have done a Greek text search on (say) epi and now I have, uh, 890 hits. I have split my frame into two side-by-side panes, GNT-T on the left and NRSV on the right. I want to scan through those 890 hits of my Greek search but I want to look for the verses that mention "earth". (This is the same as doing the original search for "epi ges", but let's say I didn't remember the word for earth.) My challenge is to skim through 890 hits to see the 6 ones that interest me. (This is a contrived example, I realize. There may be some way to do this. Although I tried making a new search text using the NRSV with [LINK GNT-T] <and> "earth" but that complains because the languages don't match.)

But that is beside the point. From my perspective (still something of a newbie) I have a great mass of text that I found by carefully handcrafting a search. I know that my answer is somewhere on this page, staring me in the face, but I can't see it yet. From my use of other common programs (web browsers, word processors, etc.) I am conditioned to do "Edit | Find" at this point to *see* what I have already *found*. I could open a new search pane and refine my search, yes -- but to me at least, that is conceptually a new task, which I am psychologically reluctant to begin until I've concluded my first task. YMMV, of course.

Perhaps after I've done a few hundred more complex searches it will so intuitive to create a new search that I will wonder why I ever wanted "find in these results." Maybe sooner. But until I reach that point, I still want one.

[jpkang]

From my use of other common programs (web browsers, word processors, etc.) I am conditioned to do "Edit | Find" at this point to *see* what I have already *found*. I could open a new search pane and refine my search, yes -- but to me at least, that is conceptually a new task, which I am psychologically reluctant to begin until I've concluded my first task.

This seems to be the trade-off between the precision in querying that structured data allows (and the discipline and learning curve it takes to use it) and the convenience of something that is almost exclusively presentation-oriented like HTML. So... is it possible to have the best of both worlds?

The dilemma is analogous to Spotlight, come to think of it; a pain to fine-tune searches, and in its current implementation, it returns data like a firehose... In this case, I'd like to be able to have deep (metadata, search-in-file) and shallow (filenames only) search presets, which is sort of like what Luke wants.

p.s. I did suggest to Luke, as one quick-and-dirty solution, that he copy and paste the search results text into a word processor where he can then do the kind of free-form, uncontextualized search he's after.

[David Lang]

Luke,

Essentially, you CAN do what you're wanting to do with Texts, and you were very close to the mark in your attempt. You just want to use the Contents command rather than the Link command. In other words, after you duplicate the window (or tab) containing your GNT-T search and switch the search text to NRSV, you can enter "earth <AND> [CONTENTS GNT-T]" to find the 80 occurrences where epi appears in the GNT-T and "earth" appears in the NRSV.

Again, I understand this is not as easy as doing a simple string search on the contents of a web page, and your idea certainly has merit. We already do this in User Tool/User Note Edit windows with the Find/Replace command. But in Edit windows you're dealing with plain, unindexed text, while in the Search window we handle text a little differently. Perhaps the Find/Replace dialog could be implemented for text in a window pane in a future rev, but since I'm not the programmer, I don't know how hard that would be.

My point is simply that in the meantime you can do what you're wanting to do using the Contents command, and such searches apply all the power of Accordance to the text you're searching, such as counting the number of hits, giving statistical details, enabling wildcard searches, etc.--all of which would not be possible with a simple string search.

I hope this helps.

[Jim Appleby]

As a basic user (at this point), often all I want to do is search for a brief pericope, when I already know what chapter or book it is in.

Because I am as resistant to change as most human beings, I really just want to be able to find something within the constraints of what I'm already looking at.

Right now, the easiest way to do this seems to be to copy the text out of accordance and copy it into pretty well any other text app and do a basic search - because I haven't been 'in the industry' long enough to really want powerful multi-step searches. This isn't such a pain, but it sure would be nice to do it in my fancy accordance program, rather than textedit.

[Helen Brown]

You can also use the RANGE command to search quickly within a range, for example: [RANGE Gen] <AND> Abram.

[Jim Appleby]

Thanks Helen. This does exactly what I was wanting.

I'd still like to advocate a more simple search solution (e.g. command-f searching within the currently displayed text). Here are the reasons:

OS X Consistency: Command-F usually allows a search within the active page of an application (e.g. text programs, html and other code editors, browsers)

Intuitiveness: Currently (unless changed, I'd assume) a new user who wants to find something hits command-f, types Abram, and finds results in every bible and tool and text in the program. Likely - at first - a person just wants to find a story in Genesis in their preferred translation.

Speed: If most users are like me, and not used to coding with <> or ; or ' " or [] symbols, then even after they have learned how to do a range search (which must be learned specifically for accordance), lifting the hands away from the keyboard to find them slows one down.

I know from the quality of your software that you've devoted plenty of time to thinking various aspects through. This would be an improvement, especially if it was done so that old users could make a choice about how command-f routes.

Jim.

[Helen Brown]

The search I suggested is simply shift-command R, enter the reference, shift-command A, and enter the search. No need to remember brackets or use the menus at all.