I know most people in here are users of Devonthink. I was wondering if any of you have tried NVIVO, a professional research tool for annotating and aggregating data. How are the two different? In what contexts one can preferably use NVIVO over Devonthink, and vise versa?

Well, one difference is that NVIVO costs almost 8 times the top of the line edition of DEVONthink.

NVivo is structured mainly for qualitative data analysis, or QDA. It is useful as a research tool in many fields, and specializes in "coding" text, images, audio, social media, and so on. As NVivo's site says "NVivo is software that supports qualitative and mixed methods research. It’s designed to help you organize, analyze and find insights in unstructured, or qualitative data like: interviews, open-ended survey responses, articles, social media and web content."

DEVONthink is far less specialized. While it is certainly used by researchers, it is actually not all that useful for QDA because its annotation features are mainly the same as macOS's annotation features in Preview. I.e., very limited. DEVONthink has also gone through a lot of issues since Sierra messed up PDFKit -- the basis for DEVONthink handling of PDFs. DEVONthink is focused on document management, primarily, and secondarily on document creation, note taking, and annotation. There is nothing that DEVONthink does in those secondary areas that is any better than the vanilla apps that come with macOS, and DEVONthink is far less adept at document creation and editing than the variety of specialized apps available on macOS. DEVONthink keeps a concordance of words and metadata properties (tags, PDF properties, etc.) for all documents that it knows about in its databases, and uses that concordance to suggest matches, similar documents, and provide speedy search. Back years ago (a decade) when DEVONthink first came out, these concordance features were call DEVONthink's "AI". It's sort of a quaint term today, in my opinion, DEVONthink's "AI" is no where near real AI.

Anyway, you can get demos of each product, NVivo and DEVONthink, and the developers are pretty helpful. If you have no idea what QDA is, then get a DEVONthink demo. Buying NVivo for document management would be like buying 5-ton van to pick up the groceries.


Dellu wrote:

I know most people in here are users of Devonthink. I was wondering if
any of you have tried NVIVO, a professional research tool for annotating
and aggregating data. How are the two different? In what contexts one
can preferably use NVIVO over Devonthink, and vise versa?

This is a broad and brief answer. In particular, I have not used NVIVO, merely read about it. NVIVO appears to be primarily a tool for analysing qualitative data; I do not know what methods it uses to do this.

DevonThink, which I have used for several years, is primarily a tool for storing, searching for and finding relations between large numbers of documents. It uses concordances for these purposes, but that is as far as its analytical features go.

Hugh wrote:

This is a broad and brief answer. In particular, I have not used NVIVO,
merely read about it. NVIVO appears to be primarily a tool for analysing
qualitative data; I do not know what methods it uses to do this.

DevonThink, which I have used for several years, is primarily a tool for
storing, searching for and finding relations between large numbers of
documents. It uses concordances for these purposes, but that is as far
as its analytical features go.

Don't bother with my answer - Paul's is much fuller!

Thank you guys.
I just tried NVIVO today. My university has a license to it. So, basically, I ge it for free. I was just wondering if it can do better than Devonthink+Tinderbox of identifying relationships among snippets of ideas.

This concept of coding seems very important for NVIVO. I have never used this kind of research tool, even if I have been around the academics for a while now. Look at some tutorials, the application seems to target social scientists (interviews and primary data like social networks). I don't do that kind of stuff. I am in theoretical linguistics. What we do is much of similar to what chemists and Biologists do.

NVIVO has similar word concordance features: more of like the Devontagent, actually. It also seems to work fine with the OneNote. That being interesting, I find it cumbersome for reading pdf files. Longer pdf documents are hard to navigate. The coding also needs a lot of manual work. I am wondering why the coding is better or different than tagging.

Paul's answer is to the point.
I would like to see if sb could comment on how the combination of Devonthink and TB would compare with the combination of NVIVO and Onenote.

I respect DEVONthink, and use it to hold my reference databases, but the real attraction is the powerful search function, complete with hit highlighting and buttons for jumping from hit to hit. I don't use any of the other features, because as others have already remarked, there are much better options available for fast, efficient synching.

But for searching rapidly through very large numbers of documents, DEVONthink is optimal.

Having said that, dtSearch is even more optimal, despite its now rather old-fashioned "look".

Finding really good search engines is a major challenge nowadays. I'm always on the lookout for apps with drill-down search functions, but the number seems if anything to be shrinking.

MadaboutDana wrote:

I respect DEVONthink, and use it to hold my reference databases, but the
real attraction is the powerful search function, complete with hit
highlighting and buttons for jumping from hit to hit. I don't use any of
the other features, because as others have already remarked, there are
much better options available for fast, efficient synching.

But for searching rapidly through very large numbers of documents,
DEVONthink is optimal.

Having said that, dtSearch is even more optimal, despite its now rather
old-fashioned "look".

Finding really good search engines is a major challenge nowadays. I'm
always on the lookout for apps with drill-down search functions, but the
number seems if anything to be shrinking.

yes, dtSEarch is powerful engine. I have used it for a while. But, I find the proximity search in Devonthink not so satisfactory. It gets fooled by repeated same term.

'active' NEAR/5 'passive' in Devonthink finds phrases like '' active, active''. It might be a bug. But, I find that annoying.

But, for me, for searching, I find Foxtrot Pro the best in its class. the proximity searches, the previews; the searching within searches...I find it to be the most complete searching software.


I am now playing with Atlas Ti and MaxQDA. Digging a bit, these two applications are much better than NVIVO for coding. The reason why I am looking into this genre is because of the coding (tagging the paragraph; or the bullet point; or the just the sentence). No other appliation can do that: not even Tinderbox. Both TB and Devonthink tag files. They cannot tag paragraphs. Onenote in Windows can link paragraphs to paragraphs (as well as tagging). Onenote of the mac cannot do that. That is why I am looking into this class of softwares.

Dellu wrote:

The reason why I am
looking into this genre is because of the coding (tagging the paragraph;
or the bullet point; or the just the sentence). No other appliation can
do that: not even Tinderbox. Both TB and Devonthink tag files. They
cannot tag paragraphs.

Curio can tag just about anything, if you need a solution for the Mac.

If you are a Tinderbox user you might want to "explode" your text by paragraph, i.e. split a single note into a set of notes divided by a recognisable character

Curio can tag whole "figures" (text boxes, images, shapes) -- but not bits of figures -- words in text, etc. QDA apps specialize in tagging the bits -- a level of detail not common outside dedicated QDA solutions.

@Prion wrote

Curio can tag just about anything, if you need a solution for the Mac

Thank you Paul. That is right. QDA can do more details: and the concept map of the tags. But, most important to me is not the concept map of the tags themselves: rather the concept map of the quotations.

look at this video to see how the concept mapping of quotations alongside the tags work in Atlas: https://www.youtube.com/watch?v=O_Z84L_J8sk


But, I am really in conflict right now. Atlas Ti is very attractive for the concept mapping. But, Maxqda is much more efficient in many other areas.

@Prion:
The problem with the exploding of text in Tinderbox is that it breaks the flow. If you have a longer note: and break it down into separate notes, really, the flow gets broken. Furthermore, pdf files cannot be annotated and tagged in Tinderbox. Only text files can live inside TB.

You have to read, the pdf in a pdf reader, write a summer or note of the article-->export the note to Tinderbox for processing.
IN Atlast, everying is just there. If you add more annotations on the PDF, you don't have to export again.

I think outliners that support tagging like Outlinely and Dynalist are excellent for this purpose. It is a bit of a compromise in that you don't tag words or documents but paragraphs (or items) - but you can break up larger blocks of text into sub-items and then tag each individually. What is nice is filtering for tags preserves the document structure.

Luhmann wrote:

I think outliners that support tagging like Outlinely and Dynalist are
excellent for this purpose. It is a bit of a compromise in that you
don't tag words or documents but paragraphs (or items) - but you can
break up larger blocks of text into sub-items and then tag each
individually. What is nice is filtering for tags preserves the document
structure.

The problem with outline softwares is the exact opposite of database softwares like TB and Devonthink. The outliners can filter within a file; but, they cannot filter across files. I am not specifically familiar with Outlinely or Dynalist you mentioned. But, from my experience, outlining softwares work within are limited within a file. If you have to take advantage of their tagging and filtering features, you are going to meed merge hundreds of files: as you need to break down in Tinderbox and Devonthink.

Not an issue for workflowy which keeps all files as one big outline, and the others can easily handle large outlines as well (even if they encourage multiple files).

And if you get Dynalist pro I think you can see keywords across files, but I don't have pro so I can't test it out...

OK, I was able to test this just now. While pro is required to see a list of keywords across all documents, you do not need pro to search keywords across documents.

Luhmann wrote:

And if you get Dynalist pro I think you can see keywords across files,
but I don't have pro so I can't test it out...

Sorry to be late to the party.
Thanks for the heads up on the foxtrot search. that's been useful!
Marginnote pro will export to devonthink. It's at the right price to get the best of annotations and content selection into devonthink at a snippet basis that makes the search of devonthink and the "AI" really work! A few fold cheaper than atlas.ti etc.

hope theres an avenue to explore?!

bigspud wrote:

Sorry to be late to the party.
Thanks for the heads up on the foxtrot search. that's been useful!
Marginnote pro will export to devonthink. It's at the right price to get
the best of annotations and content selection into devonthink at a
snippet basis that makes the search of devonthink and the "AI" really
work! A few fold cheaper than atlas.ti etc.

hope theres an avenue to explore?!

Marginnote is very pleasing application for reading PDF files. I also find the export capabilities very interesing. I tried to export my annotations, it breaks each annotation into a separate RTDF file and export it into a single folder. This exactly like the workflow we used to have Sente. The system in Sente was slighly better because the pieces of notes where not numbered; We used to assign titles which summerize the main concept of the annotation.

Marginnote still doesn't fix the break between the reading, and tagging in one hand, note taking and taking on the other.

Assume you are reading a stricking passage in the book. You want to quote that paragaraph in your future writing. What do you do in Marginnotes?
You highlight it; write a comment....that is it. You cannot tag it. Unless you remember the words or phrases, there is little chance that you will pick that paragarph when you need it.

In QDA applications, it is very simple. you assign a code (tag) to that pagagraph by the topic. If you want to quote the pagaraph when you are writing, for example, about "algea", you code the paragraph by that term. You don't need to search it when you need it. You just pick the code.

Coding (tagging) solves the weakenesses of SEARCHES. Search requires knowning the exact terminology; the exact English words. Search doesn't know conceptual connections. Codes (tags) can register and keep conceptual connections because you are putting them thinking about the connections.

I would have said Marginnote solved the poblem if I can tag the paragraph in Marginnote; and that Devonthink understands and picks the tags.
But, it is possible to develop some internal techonlogy to fix the break. One strategy is to use the system in the BEAR application, for example. You just put hash before you keywords. Then, after exporting, use some script to translate those hashed words into tags. We have been expreimenting with this kind of systme in Tinderbox forum. I also tried it with Day One, Bear and MacJournal: https://dellu.wordpress.com/2017/07/02/switching-from-day-one-to-macjournal/

As I stated earlier, while Bear and other text editors only allow you to attach tags to individual documents, outliner software such as Worfklowy, Dynalist, Outlinely, and Mudu all allow you to attach tags to individual outline items and to filter the outline for those tags. If you use outliners that export your annotations as single plaintext files, such as GoodReader, PDF Expert, etc. then you can simply copy and paste these into your outliner of choice. This has the added advantage of letting you use the outliner functions to organize the text outline to reflect the structure of the paper. (If you highlight the sub-headings this makes it much easier to do so.)

It looks like MarginNote can even export into OPML format, which would simplify the whole process?

Very strange. MarginNote only supports the proprietary Omni Outliner format, not the open OPML format supported by many other apps. Seems easier to use an outliner that exports as plain text than this...

Luhmann wrote:

It looks like MarginNote can even export into OPML format, which would
simplify the whole process?

I just tried within Outlinely. It filters only within the note. The design works only under the assumption that one would put multiple tags within a single extended note.

As I have mentioned before, this is useless to organize information of larger amount. Useless.

Outlinely has in-note taggin.

IN-Note tagging: You cannot compare the tags (ideas) you collected from one book with another book. This is even more primitive form of tagging than the one I have seen in Bear. Bear can at leas search the notes that contain that specific tag across: and put them together. One can use further searching to look into each of the notes. BEAR uses the File-based tagging; just like the rest of mac applications.

The IN-note tagging is for long notes. It is a promising direction: but, it lacks the capacity to link ideas across notes. The QDA applications unify the two tagging systems such that ideas you tag in Book1 will still be visible or comparable with the taggs you put in Book2....

Outlinely can search for tags across documents using the "search and navigate window." Dynalist can search in a single document or across documents. Workflowy does not require you to separate outlines into separate documents. (In fact, it isn't necessary in the other apps either - they can all handle very large outlines without suffering.) The point I was making though is that you have greater granularity with an outliner document as you can filter the document to see only those items which are tagged. This is not possible in Bear or similar apps.

only because it doesn't suit the forum title,

has anyone had a dive into omnity.io?

may be useful to you data miners!

best,
wade