Why confuse the issue?
Anyone really familiar with the vast collection of Apache projects knows the acronym SOFA already belongs to a completely different one - UIMA, where it stands for "Subject of Analysis". Now you'll have people thinking it has something to do with CouchDB.
(Of course, it's quite possible that you'd take a lot of unstructured text data, store it in CouchDB, and then extract parts of it into a SOFA for NLP processing with UIMA. What larks, Pip!)