You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
to Annif Users
Hi all,
I'm very interested in the subject inclusion/exclusion functionality available in Annif 1.4+. However, I'd like to ask about its behaviour when using ensembles. In particular, I know you cannot normally combine different vocabularies within an ensemble.
However, with this new functionality, is it possible to have a large vocabulary, and two separate backends that each specify a *different* subset of included subjects, and then combine their results using an ensemble (since the unfiltered vocabulary is the same)?
A use case example would be something like LCSH/LCNAF, which is often expressed and applied as a very large single vocabulary. One backend could be specified with the LCSH subset, and another backend with the LCNAF subset, with the results combined into the same label space. This would presumably reduce the resources needed to train each subset, since the respective backends only "see" the filtered subset.
My apologies if I've missed or misunderstood the documentation on this; I wanted to ask here before I started experimenting and started getting confused.
Thanks,
MJ
Osma Suominen
unread,
Jan 7, 2026, 4:06:20 AM (10 days ago) Jan 7
Reply to author
Sign in to reply to author
Forward
Sign in to forward
Delete
You do not have permission to delete messages in this group
Copy link
Report message
Show original message
Either email addresses are anonymous for this group or you need the view member email addresses permission to view the original message
If you want to try it out with LCSH+LCNAF, you should define a single
vocabulary (maybe called lcsh_lcnaf or something along those lines) that
both the ensemble and the individual projects use. Then set
exclude/include rules that narrow down the vocabulary for the individual
projects. See the above comment for a similar configuration for the GND
vocabulary.
Please do report back on how this worked for you! This is a new feature
so we don't have much experience with it yet.