Recently , in a blog post DT put a link to a tagging podcast. This prompted me to mention something I have found.
I have been using DT pro for 3 to 4 years. A lifeline for me given the amount of info coming in to manage and not wanting to carry around luggage full of documents.
When I started I read the very good book by Mr Kissell on DT. He said you are either a tagger or a grouper. Which is a good starting point but it is not completely correct. As I recently discovered tagging is a means of developing a separate and supplementary method of organisation of information as well as grouping. Not just for searching purposes but for viewing purposes in DT pro.
An example, is the best way to illustrate the point. I’m involved in environmental cases where there are multiple parties and multiple expert briefs on multiple topics such as marine ecology, landscape, recreation and navigation.
A logical grouping is based on which party the evidence if for. So the applicant’s evidence is one group. The regulator’s another. The PDFs briefs are therefore based on the names of the parties. However the usefulness of that grouping is not great when preparing for cross-examination or comparing and contrasting expert opinions on a topic.
If each expert brief from all parties is tagged according to topic or discipline you can then use DT to view easily all of the evidence on a common topic. This is achieved by going to the pane view for tags in DT and selecting the tag view option. Then you will see all of the tags on the right handside . If you then select a tag topic all of the documents fitting that topic come into the document viewer.
Having multiple ways of grouping information and accessing information is fantastic when confronting a waterfall of information and you have a range of ways you want to organise it.
Outside of DEVONthink Professional Office, I’m neither a grouper nor a tagger. I worked quite well for many years relying primarily on information-rich titles (YYYYMMDD + keywords) and searching (HoudahSpot has been wonderful). In DEVONthink, though, I have found the way that the search is set up isn’t very compatible with my system, which has generally relied upon using some kind of search grammar in the search field (think Evernote or BBEdit).
I could see myself using tags in DTPO, especially because I really like how well it handles parents and children. However, there is one drawback: DEVONthink To Go. On the iPad, as far as I know, you really cannot do much with tags. I’ve found the search on DTTG to be anemic, and so I need somethig in order to navigate thousands of files. Groups work wonderfully.
I’m not a big fan of folders (the closest OSX analogy to groups), but one nice thing about DT is that it allows for replicants, and this makes all the difference. I can file away something into one topic (maybe because it is related to a particular working group) and I can replicate the file in another group related to something else (maybe a topic shared with another working group). Groups aren’t terribly different than tags when they are used this way (this makes sense, because gropus were tags in the past, and can still be used that way).
Another nice thing about DT is that the AI works according to groups. It doesn’t recognize tags, so if you want to take advantage of the AI features, you’ll need to use groups wisely.
All of this is not to say that I am against tags. But, whether you are a grouper or a tagger might not depend so much on your preferred method of organizing as it does on the particular tasks you are doing with DT. If you are planning to use the AI features or the iPad, groups are going to come in very handy. If I only worked on the Mac, I’d probably have a lot fewer grops and a lot more tags.
These are very interesting use cases – thanks to @john_massen and @FROBGOBLIN.
DEVONthink treats tags as groups – so tagging results in a alternate hierarchy for your data. A different “slice” or “view”. I think the thing to consider before investing a lot of time in any kind of naming scheme, or folder/group hierarchy (including tagging) is whether the work to create and maintain that scheme or hierarchy will pay back over time.
When OpenMeta tagging first came along, and then again when DEVONthink first introduced tagging, I spent a lot of time very carefully nuturing a complex taxonomy that I believed would have tremendous benefits. It never did. The tagging never supplanted the formal group structures that I use in my work, personal, and research databases. I’m down to a handful of regularly used tags that help with workflow: “readme”, “now due”, “WIP”, “done”.
I’m not suggesting tagging is, or is not, useful as a practice. Just suggesting “be thoughtful about what you can actually accomplish”.
Can long-time experts in DT answer a couple questions related to this post?
What, in your mind, is the difference between A) using tags for parties and groups for topics, vs. B) tags for topics and groups for parties? Does one allow you to do something (more easily, more quickly, less often) than the other?
Is it more efficient to use both See Also and Classify, or is just using one of them as efficient as using both?
Are See Also and Classify used for different things, or are they just different ways of doing the same thing? I understand how each works: Classify is based off user-defined groupings and suggests groups, whereas See Also suggests individual documents regardless of their grouping. But I wonder if there are implications that aren’t obvious. For example, is See Also best used for documents that you already have read and know their content, whereas Classify is best used for unread/newly-imported documents? Is Classify best for a first-pass general sorting, whereas See Also allows you to look for anomalies or uncommon connections?
Does the length, complexity, overlap of documents make a big difference for the AI?
I realize the answer can always be “It depends on what you want to use it for/how you use it”, but to make the most of DT’s AI, how would you set it up?
In my mind - no semantic difference. (Though I not 100% clear on the meaning of “parties” and “topics” in this context.) There is a functional difference between tags (children of the Tags group) and “normal” groups: if you delete a document from a normal group, and there are no replicants in other normal groups, then the document and its replicants in Tags are deleted. But the tag is not deleted. (DEVONthink assumes you want the tag to persist for future use.)
I think your general understanding of See Also and Classify is correct – though they are definitely not “different was of doing the same thing”. You description in “Classify is based off…” is more accurate. The replicants stored in “Tags” are ignored for See Also and Classify.
The AI works better when there are more samples (documents) – but it’s a long-standing question here if smaller files are better than larger. I don’t know the answer, but I know there is a lot of opinion posted here over time on that topic
Re the difference between “parties” and “topics” in choice between using tags or groups:
The Classify assistant works best when groups hold topical content – assuming that a topic is likely to involve documents that use similar terms and associations of terms. It is likely to be less useful for groups that are non-topical and hold documents that vary widely in the terms and associations of terms that are used.
So, if I wish to use the Classify assistant to make filing suggestions, I would use groups to hold topics and tags to denote “parties” (I assume you mean the names of individuals).
On the other hand, if I’ve created a group to hold documents specific to a person, such as Mary R. Smith, and I can identify the person by a glance at each document (or by doing a search for her name in the case of batches to be filed), it’s a simple matter to file by name of individual. The Classify assistant would not be needed (and wouldn’t work well in this case).
I organize by topic in my research collections, but by category/year in my financial database. Classify works well in the research collections, would be useless (but not needed) in the financial collection. Works for me.
Note: I frequently collect scans, most of them related to financial affairs such as banking statements, bills and receipts into a database created for that purpose. I used the full Search window to do searches designed to segregate the searchable PDFs by source, e.g, name of bank, checking account or savings account or IRA or investment account. I’ve got a total of 33 such smart groups. After a scanning session I can very quickly file items into the organization of my financial database. Often, there’s stuff left over in the incoming scans database that didn’t get listed in a smart group – usually, only a few items that require attention for filing decisions. Hey, I’m lazy.
I agree , I am only using about 7 tags. Organising workflow is an evolving art but my goal is advocacy without paper and DT through its search and viewing options and general functionality gets me closest to this.
“parties” in this context means the people who have status in the proceeding to present their evidence and case. These include the person seeking consent , the regulator, affected people and interest groups.
Very interesting point though that the choice of what is the group and what is the tag affects the efficacy of the AI in “see also” and “classify”. That is going to make me do a bit of a rethink for the future when developing a database for the next case.
Thank you for your posts John. I have followed them with interest. I have written quite a bit on these forums about using DT for legal people (being one myself).
You might find the approach to tagging and folders I discuss in this thread pertinent to what you do:
[url]Put up Example page]
The problem I found tagging whole documents, is that its all very well to locate the documents, which say are tagged as ‘Regulation 66’, but each document may have 10s of pages, and fifty documents may be tagged as ‘Regulation 66’. I wanted to see only those portions of the documents that related to ‘Regulation 66’ and to be able to extract them for the purposes of preparing legal briefs and witness summaries.
To meet this need I created an annotation script (which is generally useful to other people) but which had its genesis in seeing some windows based legal tools such as Lexisnexis CaseMap.
There are a number of attorneys, solicitors, barrister, advocates, paralegals and legal academics (different names depending on where you are in the world) using DT and it seems to me it might be worth starting a thread especially for lawyer’s workflows.
Hi Frederiko , yes I have followed your posts also with interest. You have way more knowledge then I do as I have not delved into scripting. Thanks for the posts , it helps a lot.
On your linked post you say
Thats what I do for tagging.
The point of my latest post was that I use tagging to quickly access some groups of documents in addition to grouping. To rely on a computer rather than traditional paper in a courtroom requires me to immerse myself in every feature of DT so that I am not slower than anybody else to access information.
I , like you, use the wiki link feature a lot. And I use it to prepare my cross exam. This tends to be topic based also rather then scripted questions .I try to cross exam using my computer. Because of this I bring up a document in cross examination relevant to the question using the command right click function on the link. The whole wiki link area needs more exploration by me.
Your particular work appears to involve multiple briefs all with pieces of information relevant to sub topics e.g misrepresentation and you need to extract related parts from multiple briefs. I don’t have that issue so much as all the evidence I deal with tends to be expert and topic based. I do some civil litigation but not much but can readily see for e discovery and large cases the DT functionality would help enormously. In fact I think DT is a great tool for lawyers. Both my sons are at law school and use it and my wife is learning it and I am going to introduce it to the lawyer I work closely with who is my understudy.
But it really does involve some effort to learn at the start . Unlike some people in the forum when working I stay in DT and DA pretty much all the time rather than using finder safari and mail. It may be too restrictive for some people’s taste but it was the way I become comfortable with the product.
I would be interested in a lawyers focussed post also as it may be a way to access info particularly relevant to our world. I had one lawyer a year or two ago from Brisbane want to exchange info following a post but I couldn’t follow through because of work commitments. And thats an issue because on long cause matters ( i.e. trials) I am pretty much 100% focussed on the case and not posting to DT. You must be the same.
Incidentally I am thinking of taking one step further towards the mobility grail and get a new macbook to take to Court on procedural arguments and short cause appeals. Do you know if it handles big pfd’s operating in DT well?