If you work in (geo)data, you most likely have come across the term “linked data”, see for example LINDAS, the linked data service by the federal government. If you’d like to dive into the topic a bit, this blog post (in German) by Stefan Kaufmann explains the basic idea using postal addresses as a simple area of application (machine-translated to English, and to French; the title is a bit mangled in both, as the German avoids using an umlaut, but otherwise the translations ought to work okay).
From the article:
Because, just as with spoken language, we humans are quite good at interpreting ambiguities or imprecise expressions according to their context. We don’t just read what’s written somewhere and interpret it literally; we interpret without having to think too much about what might have been meant . Unfortunately, when we write, we also assume that the other person will already know what is actually meant . And that this other person isn’t a computer that interprets what is written comparatively literally. And then things get wild.
The article adopts a mostly European perspective and doesn’t even mention other addressing conventions such as the ones used in Japan. If this interests you: A few years back, Ed Freyfogle of OpenCage talked about geocoding on the MapScaping Podcast. And some of the complexity and variety of address data are evident in this collection of address formatting rules by OpenCage.