I’m processing 3,000 pdfs & docs into about 40 groups, and a couple of things are irksome: the first is the necessity to re-adjust the view setting (I prefer Zoom to Width) for every file. It’s also frustrating that there is no Toolbar Icon available for Zoom to Width.
There are several approaches to organizing a large number of unclassified items into an organizational structure, some of which obviate most or all of the problems you have noted.
First, when you open a view window and display a PDF document, select View > Fit to Width. Thereafter, when you select a PDF it will fit the width of the text pane in the view. The text of the displayed PDF will be easer to read than if it were displayed, for example, in View > Zoom to Fit.
Depending on the screen real estate on your computer, you may wish to choose a View window that provides the largest display of your PDF documents. On my 13-inch ModBook, that’s the Vertical Split view.
Although when you are viewing a PDF in either the Vertical Split or Three Panes view the Classify button is available, the Classify button will not be of much use until you have sufficiently populated the groups in your organizational structure to allow the AI function to recognize the contextual relationships of the documents in each group.
As you are starting with about 3,000 unclassified PDFs and have established some 40 groups into which you wish to organize them, you will need to start by seeding each group with manually selected documents. At some point, the Classify button will become useful. If you have painstakingly organized content, at some point Auto-Classify will become useful. Tip: If you use a group, such as Inbox, to hold unclassified material, exclude that group from classification (in the Info panel).
You could start by selecting the first unclassified PDF, reading it or making a judgement based on the document name, then using the contextual menu option Move To to send it to the appropriate group, or drag it to the appropriate group in the Groups panel. But with a backlog of 3,000 documents, that’s a lot of work. I wouldn’t have the patience to do that.
One approach that may work well, at least for some of your group classifications and for some of your PDFs, might be use of a search for one or more terms. Note that the search results list is ranked. If, for example, one of your groups is intended to include articles about invasive species introduced into an ecological setting, you might find that relatively highly ranked search results for “invasive species” would fit well into that group. So you’ve quickly moved a number of items, instead of doing that one by one.
Another approach, either taken first or as a supplement to search classification, would be selection of non-contiguous items (Command-click) based on judgements from the titles of articles. Of course, the more descriptive the titles are, the more useful is this approach. Just run down the list of document names, Command-clicking on each that appears to fit into a specific group. When a number of selections have been made, either Command-click on the selected items and choose the Move To contextual menu option, or drag the selected items to the appropriate group in the floating Groups panel (Tools > Show Groups).
Still another approach is to select a document and invoke See Also. That AI routine will examine the contextual usage of the terms in the viewed document, look at the contextual relationships of that document to all the others in your database, and suggest possibly similar items. That can often be useful for initial seeding of a group with related content.
Except for simple cases, such as filing highly specific and useful items like tax and expense records or correspondence, I decided long ago that there’s no perfect organizational or other tagging system that can accommodate the variety of ways in which I need to access information for different purposes.
So as time passes I find myself devoting less time and effort to overall organization of a database, but instead concentrate that time and effort more specifically for each new research/writing project I undertake, and do that more concentrated organization within the new project group, to the degree that it helps me get the job done. Most articles and books contain numerous information “elements”, some of which may be important for one purpose but not another. Fortunately, DT provides tools that are independent of organizational structure, including Search, Spelling, Context, See Also and See Selected Text. So one can mine information from a reference collection independently of – or even in spite of – its organizational structure.
Bill,
this is one of the most useful descriptions of data into DevonThink that I’ve read. it would be a terrific addition to the DT Academy tutorials. It’s a bit “lost” here, and I know that I want to revisit this advice repeatedly.
Linn