I've been meaning to test this tagger for a couple of years now ... it only seems to have GD3 as a premium database which made me a little reluctant, but I'm interested for its retagging abilities. I need to find the time.
However your question may be simple but the response is not at all simple.
First there are a lot of different opinions on classical tagging (do an advanced search on this in the forum and you will find a bunch of interesting posts. Unfortunately the metadata easily obtainable from the internet upon ripping is just not good enough, or its insufficient for even a medium sized classical collection.
I simplified a bit what I do in a recent post that you might want to skim over here.
http://yabb.jriver.com/interact/index.php?topic=102859.msg713970#msg713970, go to the end of the post where I gt a bit more specific about classical stuff
http://yabb.jriver.com/interact/index.php?topic=102859.msg717597#msg717597. Bear in mind there is no right way to do this. What is important is to have as much detail, tagged in a
consistent way, so that as your library progresses you can change your mind, and your "views" relatively easily - to avoid redoing the grunt work if you will.
Sometimes this could involve using multiple fields with the same data. You should have an idea how you want to separate different versions of the same work, and how you want to see your playback choices on the screen
One thing to really understand especially if you have a large collection is that JRiver uses the Album Name and the Album Artist as the default grouping. You say that composer is a custom tag ... actually it is not. It is easy to set up a view based on just classical music, its just common sense to have at least one of your main views based on Composer. However it depends on how you prefer to handle certain scenarios like for multiple composer albums, three versions of the same opera, deciding if you want to group the original sequence order from a CD, or do you want it filtered by work etc.
You mention the field "composition", if written to the file via this tagger, you (most of the time) can remap it by creating a custom field and import it to JRiver. My field name that I use is called [Work], but sounds like it is the same. However, some people with large collections need an "Opus" field too (like for BMV numbers etc.). If you have the complte works of a composer like Mozart, adding different versions of some of them, you might need it too. You can also stay simple, and just have a View based on composer, then album where you would not need all this detail of course.
Anyway this sort of fascinates me, so if you want to copy/paste your key classical fields with an example of how they are filled in, I could come up with a suggestion or two if that would help?
PS-highly recommend FLAC as the format of choice, especially with Classical music -- as it is the most flexible container for extended metadata and compatibility. Also, how are you handling multiple performers (soloists) -- the semi-colon is the default separator in JRiver ... Im' not sure how that is handled in the tagger you are using. As JRiver is not relational, I'd be interested in knowing how Music chi handles the field [Instrument] for example. Personally I'm not a fan of CUE files either, if you use them you might want to ask yourself why you do ... for me they only serve (maybe) for burning cds, but that's another discussion.