I am a Together user that has been trying to make the jump to DevonThink Pro. However, some functionality that I use frequently in Together is the ability to quickly do ad hoc tag searches. For example, I select Tags as the search type and in the main search box I enter okie california tent. As expected, a list of results is produced that is the intersection of those tags.
Unless I am missing something really obvious, in DevonThink Pro, I have to go to Tags view and scroll around in a few hundred tags. This is simply not viable when I am trying to be productive.
I could create smart folder or do an advanced search. Again, this isn’t really viable and makes something that should be very simple into something that is overly time consuming.
Am I overlooking something? It would seem intuitive that Tags would be an option in the drop down list in the search box.
Tags can be included in search criteria using the Advanced button in the full Search window. (I do all my searches in the full Search window, as it provides a number of options not available in the little Toolbar query.)
In the Advanced search panel, you can add multiple instances of a predicate by holding down the Option key while clicking on the "+ button. So it’s easy to set up a search for items that have tags “A”, “D” and “M”, for example. Or that have tags “A” and “D”, but NOT “M”. (This is more powerful than just a drop-down list of tags.)
Of course, the full Search window would then allow further filtering, such as Flagged items that contain a specific phrase.
Tag search in the Toolbox query? I’m not the developer, so don’t know whether that’s possible or planned. As a feature request, it would be more likely to get the developer’s attention if supported by other users.
One can already view items that have a certain tag by selecting that tag in the Tags view, or by opening the corresponding tag group within the Tags group.
But such a Tags search option feature in the Toolbox query would be much less powerful than the approach I suggested in the full Search window. And, especially if a query is one that I might repeat, any search set up in the full Search window can automatically create a smart group simply by clicking the “+” symbol to the right of the query field and naming the new smart group.
Bill, thank you for your response. It is obvious that we use DT in different ways. I very much appreciate the power available. However, there are many times when I will gladly trade power for speed.
When working on a book, I often need to blast through several searches when looking for resources.
I may do the following tag searches:
okie AND california AND tent
okie AND california AND shack
migrant AND tent
lange AND migrant
I would want to do those searches as quickly as possible. I would replicate promising material to whatever project folder I am working in. There is no need to save the searches or make smart folders for them - they are simply ad hoc, throwaway queries.
I like DEVONthink and I hope that this functionality will evolve eventually.
Haven’t moved my car since a heavy snow on 26 December, and haven’t bothered to shave since then. My beard is white. As it should be, at 81.
But the snow has finally melted to the point that I’ll be mobile tomorrow, so I’ll shave in the morning.
Our workflows appear to be somewhat similar when working on a project, as I’ll do a lot of searches for useful material and replicate it to a project group. Except that I almost never tag anything in advance, but rely on text searches, See Also and my organizational structure to collect useful material. At that point I may create a few tags to help me remember which items are most relevant for a particular section of the project – and will wipe them from the database when the project is finished.
Many years ago I ran a computer information center before the existence of full text indexing of documents. I developed an abiding suspicion of the logical sufficiency or consistency of application of keyword or tag terms to adequately describe documents without spending a lot of time and effort, and even then the results were often unsatisfactory.
I work with topical databases each of which serves a particular interest or need, so there’s an initial decision about which database is to receive new content. I use groups to hold items that have related content, but I don’t spend a lot of time refining my group structures. Often. I happily dump tens, hundreds or thousands of documents into a database without assigning a single tag up front. Groups and tags are useful tools, as are the many annotation notes I make within a database. But I think the ROI from using DEVONthink to help me find needed information using searches and See Also is much better than the ROI of tagging every new item as it is added. Others are free to disagree.
Bill, a lot of my research material is comprised of photographs (usually jpeg files) that I have tagged. Text searching will not return the desired results. A work around would be to put keywords in the filename but this will quickly become unwieldy.
A hierarchical approach is also insufficient.
Tagging is the best solution I have found so far. I can tag a photograph with key elements like dustbowl, car, migrant, okie, lange, family, children.
By the way, my beard is getting closer to white than grey.
It could be that the ROI from using search and See Also is much better because the tools for searching on tags is so weak. I’ve beat this horse to death, and while there have been some improvements when using an advanced search for tags, the tag retrieval capabilities are still woefully inadequate.
As is, tags are only marginally usefully in DEVONthink if one restricts the total number of tags, or only uses tags for specific projects as Bill suggested. Many posters here appear to only use (Group) tags as a means for classification, rather than a tool for document retrieval. The Tags view is nearly useless when there are many tags in use in the database, but would have much greater utility if only the Tags view worked as a tag cloud to show the intersection of the selected tags.
I’ve also been very vocal that the only reason that I index my databases is so that I can use Punakea for some decent OpenMeta searching capability. At one time I even used Together in conjunction with DEVONthink, indexing the Together folder hierarchy into a DEVONthink database (not worth the aggravation, in case anyone is wondering). I’d much rather have most of my documents contained in the database rather than indexed, but until DEVONthink gets some useful tag search and retrieval tools, I’m sticking with indexing.
In other words, put me down for a big PLUS 1 for Greybeard’s request.
I forgot to menton this to Greybeard earlier-you might want to check out Ammonite. It is a very functional tag search utility that will do exactly what you want, searching inside the DEVONthink databases. While I’ve paid for Ammonite, I’m not a fan of the dark, HUD-style interface. If that’s not an issue for you, it may be exactly what you need for now.
Greg, you are on target about the utility of tagging for filetypes that don’t contain text. There are fewer than a hundred such files (photos and videos) in my Main database holding about 30,000 documents. But I do tag the non-text files, as they always have a specific context related to other documents in my database and I want to mark them for their purposes.
That database contains thousands of scientific papers, news articles, summaries or excerpts of laws and regulations (U.S. and EU, primarily), case histories and policy papers related to environmental topics (my professional interest for many years). References span sizes from abstracts to book length. It also contains thousands of my own notes, which range from a few words to hundreds of words and are typically rich in links to other documents. I’ve been building and updating that database for more than 10 years.
Long ago, I gave up on the utility to me of the effort of trying to “mark” the importance or content of a reference document by using a set of keyword or tag descriptors. The simple reason is that my interest in a particular paper or book is usually contextual, that is, related to the research project at hand. So my interest in a particular reference and my focus on specific content within it will often vary. To put it simply, keywords or tags previously assigned are likely not to be relevant to the importance of that reference for another project. The time and effort I had made to “mark” that reference wouldn’t be repaid.
Instead, I do spend time and effort – sometimes a great deal of both – in creating rich text notes within the database. When I’m reading a paper or report I may create such a note to enter summaries and comments, often including references and links to other documents on the same or a related topic. I do my draft writing within the database, and those notes provide both the grist of and the source citations for those drafts. Of course, all my references and notes are searchable and so at my fingertips, including the ability to invoke See Related Text (which operates like See Also) on a selection of text (great for breaking writer’s block).
Sometimes writing is fun, but at other times it’s just hard work. In either case, I appreciate the help that DEVONthink provides.
I’d also very much like to be able to search tags from the search bar.
One of my primary uses of DTPO is as a research and readings database for a class I teach on food systems. I organize this database using topical tags: organic agriculture, vegetarianism, nutrition, etc. I also use a small number of important metadata tags, such as for academic vs. popular vs. government sources. So if I want to see all and only the academic pieces on vegetarianism, I have to do an advanced search. That’s not so bad, but it’s not consistent with doing a name search.
More importantly, my students access the database through DTPO’s web interface. Since advanced searching isn’t available in the web interface, there’s basically no way for them to find all and only the academic pieces on vegetarianism.
I probably shouldn’t weigh in on this since I am obviously biased (having come from Ironic Software and having been part of the birth of OpenMeta (though not programatically)) - lol, but…
I not only support Greybeard’s suggestion, my horrible bias is for full blown NSMetadataQuery style support, either as Raw Queries of -literal Spotlight search structures. However this requires more plumbing than we have in place. Spotlight doesn’t index Packages so it would really be pseudo-queries that look like mdfind queries (not searching Cache files but the actual database). This would essentially be trying to replicate the Spotlight APIs in DEVONthink. While that would thrill me to death, I think we have bigger fish to fry (or deceased equines to flog ).
However, a simple “Tag:searchString” option (similar to what was provided with the OpenMeta.mdimporter for use in Spotlight searching) would be nice. Especially nice would be simple Booleans (that we never satisfactorily achieved at Ironic) like “Tag:DEVONtech” NOT “Tag:DEVONsphere” (or the like - assuming AND). Just my simple opinion.
“TAG:foo AND TAG:bar NOT TAG:baz” in the search bar to get files that have both foo and bar tags but not the baz tag would be invaluable. We all know how to use “advanced search,” but folks use DTPO for speed and efficiency, and advanced search is not. (Plus, even advanced search doe snot allow for wildcarding of tags.) And the Tags and Groups view, while helpful for flagging intersections, is unwieldy when one has lots of groups but relatively few tags. While Bill may not like tags, but many of us do, and DTPO has gone to the trouble of putting in substantial tag infrastructure already.
This is not a “wouldn’t it be nice” kind of request. Search functionality and efficiency are at the core of DTPO, and making tags at least as accessible as metadata – in the search bar – seems like a no brainer.
)