Ideally these collections are queried from an outside source. If that's available, we can simply implement a connector for it, as we do with machine translation services.
