Bless this mess...Can DTP's "AI functions" clean i

After a few months of having DTP (1.1.1 now 1.2.1) on the dock, opening it every so often, and all the while dropping stuff in it, I now have a cosmic mess. It’s quite embarrassing, but maybe others have/have similar messes.
Is there a way of using “AI functions” for cleaning up? (I’m not exactly sure what this term means. It seems to appear a lot in the forum; there’s no definitive list in the manual. The video is pretty nice, but is about a single file.)
Most, but not all, the files are linked. The db is about 870MB. There are quite a few duplicates. Despite the advice in the manual, I’ve got sub-folders and documents mixed in the groups in the hierarchy.
My own folder structure seems confusing to me because it has both projects and topics represented as groups. This is similar to pre-DTP days, when I would have a topical folder under a project (say, a book chapter). I turned to DTP because I couldn’t conceive of going through the pain and agony of opening every single document. My work is highly interdisciplinary, and I found that as I started to organize new material as topics, I lost the connection that they had with my thinking about their relation to various projects. In keeping with the manual (or maybe some advice on the forum), I tried to keep the topics as fine-grained as I could to improve the classify. This doesn’t seem to help a great deal. “Auto classify” often merely brought up the log book indicating that the item was not classified. “Autogrouping” had “story” and “identity theft” in the same grouping.
I tried keeping all new files in a special folder that I would use to syncronize, mostly the imported files, but probably also some indexed files. Somehow I wound up duplicating the files in that folder, and the duplicates appear to be breeding. I stopped using that file. I stopped putting the new files in DTP. You wouldn’t believe the mess on my desktop now.
Normally I would give up on something like DTP as being too fussy and suspiciously willing to serve procrastinative impulses. However, it seems that it is just what an interdisciplinary scholar would find useful.
Advice would be greatly appreciated. Thanks.

Suggestions

  1. Don’t create too many folders or sub-folders. Search terms will help you find relevant notes.

  2. If you know that an item belongs in two different places, use the Replicate method instead of Duplicate. A replicant is not a copy but a twin; revise one, and the other changes.

  3. Use the Information window to enter thoughts about the item’s relation to various projects. A few simple tags will do (and keep a list of those tags in the database).

  4. Keep it simple and don’t over-anticipate where the data should go. That way, early ideas of organization won’t limit you at the time of writing.