DevonThink’s artificial intelligence can be very helpful in organizing a collection of journal articles or other informational documents. To profit from this feature you may need to shift your paradigm: stop to think of DevonThink’s group hierarchy as a filing cabinet that stores your documents and start to think of it as a catalog to a filing system that is hidden from your view.
Many people think of DevonThink’s group structure as a bookcase or a filing cabinet that stores their documents, and of its tags list as a catalog to that depository. Accordingly, they treat their groups as folders corresponding to a project, author, or source, and tag their documents by topic.
It is, however, more accurate to think of DevonThink’s databases as libraries whose documents are stored in a repository hidden from view and barred to the users. The users can access the documents in the repository (but not the repository itself) by means of two catalogs: the group structure and the tags list.
Which points to an important difference between DevonThink’s group structure and the folder hierarchy displayed in OS X’s Finder: each file and each folder has a unique place in OS X’s folder hierarchy, whereas in DevonThink’s group structure you can refer to any document or group in as many places as you like.
So, DevonThink’s group structure is more like a structured index than a hierarchical table of contents and DevonThink’s groups are more like tags and less like folders than you might have thought.
DevonThink’s groups and tags are, in fact, technically the same kind of things: ‘groups are tags, and tags are groups’ as DevonThink’s help file unhelpfully used to tell us.
However, from the user’s point of view, there is an important difference between groups and tags: DevonThink’s AI can help to assign documents to groups, but not to label them with tags.
This means that for many purposes it is best to group documents by subject and use DevonThink’s tags as means to collect documents relevant to one’s projects. In this way, you can use DevonThink’s Artificial Intelligence more effectively.
Consider, for example, an ecology consultant whose database contains information about species, habitats, surveys, legislation, methodological issues, clients and so on.
If each document were pertinent to exactly one of her projects it is probably best to group them by project. However, in that case, there would be no need to use an expensive and complicated free-form database program with a steep learning curve like DevonThink. The combination of OS X’s folder hierarchy and Spotlight would probably suffice.
If, on the other hand, the database contains documents that are relevant to several projects, documents that are of general relevance but do not belong to specific projects, documents that are currently not relevant but might be relevant to some future project, and so on, grouping them by subject allows this consultant to use DevonThink’s AI both to quickly catalogue new documents, and to quickly identify documents relevant to a new project. If she tags the latter with a tag specific to that project, she can subsequently access them easily by means of that tag.
The point is: DevonThink’s AI works best when there is a discernible relationship between the contents of the documents in a group that distinguishes these documents from documents outside the group. Grouping documents by project, author, source, publication date, file name and so on, usually does not result in groups whose content is related in this way.
One more thing …
To prevent groups from cluttering your tags list check ‘Exclude Groups from Tagging’ in the Database Properties (Choose File → Database Properties → [Name of the Database] in the main menu or Data Base Properties in the contextual menu of the database in the list of open databases).
In sum, I recommend:
Think of the group structure in a DevonThink database as a catalog rather than as the place where your documents are stored
Use DevonThink’s groups to tag your documents by topic
Use DevonThink’s tags to flag documents for special purposes