In theory it can be used for anything, we just need to figure out how well it performs in MySQL and PostgreSQL - which should be possible to do with an early prototype. I am not worried about the context map building part, that can be later done in adhoc task if slow.