Considering the case of images in my database that were all saved from Facebook. These kinds of images don't even have any XMP or EXIF data! What is the best way to round up all such images in my database?
What do you mean by "round up"? Get them all in a category?
As they come from Facebook, you probably saved them all in a certain folder? You could then base your criteria on that.
IMatch automatically creates an XMP record for each file that comes into the database. This may be a very basic record if the source file has no metadata itself, but many EXIF tags etc. can be created from the image itself.
Probably a filter or data-driven category on a usually filled EXIF tag (camera make or model perhaps) should be your best bet.
Well, let's consider the most basic scenario possible. You say that an XMP record is automatically created. So, then, let's just say that the image itself, had no metadata, to begin with. What are the very first tags that IMatch will put on that file?
Just look at one of your files in the Metadata Panel using the Browser layout. This shows you all the tags ExifTool / IMatch have created for the file.
So, okay, I use the Metadata Browser on such an image, in Browser mode. It gives headings for Composite, IMatch Metadata, JFIF, JEPEG Adobe, and JPEG SOF. None of the entries under those 5 headings are for any date info whatsoever. So, assuming I would have 700 such jpeg images out of 2,500 jpegs scattered in various folder on my HD, what is the best strategy I could use in IMatch to bring just those into view inside the file window, so that I could assign those 700 image files to a category?
Perhaps this will help. I had a number of scanned images that were missing original date/time (XMP::photoshop\DateCreated\DateCreated\0), so I set up a data-driven category to locate these. Note that I included the 'Other' element; the 'Other' contained the images that were missing that particular tag.
You should be able to use a similar approach to find files that are missing a particular tag.