I have read for best results “especially when using features that rely on artificial intelligence” to limit the “number of groups in a database (say, fewer than a hundred or so), nested only a few layers deep, and with more than a few documents in each one” is the way to go.
My question, Does using multiple layers of folders have a great impact on DTs ability to search Auto Group and Auto Classify when using AI?
Also, do naming conventions of files and folders that utilize the underscore and dash impede DTs ability?
Where did you read this? Actually both the number of groups and the depth of the hierarchy are unlimited. More important is not to mix unrelated stuff in a group (e.g. databases with few groups, each containing hundreds or even thousands of documents). Related stuff should be always grouped for best results.