For example, consider the note idealism.
There, I could say I’ll be classifying Indian idealism by listing the philosophies and the philosophers under them (because there are more philosophers per ideology), while in case of German idealism, I’d be listing the philosophers first (because there are fewer philosophers per ideology, and in a sense, more philosophers than ideologies).
So we’ll need to state this upfront, so that a person trying to parse the data by reading or by code will be able to know how to do it. Now a person could ask, why do the two sections use different classification schemes? The answer is of course to be more efficient and reduce the duplication in having to say so-and-so philosopher follows the same idea X (listed before) in case of Indian idealism, or to say so and so ideologies have one philosopher each under it, in case of German or Western idealism.
So this question itself comes from the notion that classification has to be symmetric. I had this notion myself and that’s why I tried to learn the theory of classification to find the perfect way of classifying things, only to realize this topic is only formally studied in machine learning, and that comes out with highly complex systems far beyond human usefulness, involving hundreds or thousands of layers of neural networks. The note I made for this subject was classification, but the actual notes on this topic are in another vault on my phone.
So a person who deals with classified data should also understand that the system of classification is also data dependant, with no meaningful and simple universal method of classification. So one must ask the way in which it is classified (or understand that) before asking for (or reading) the classified data.