Problems lurking for any creator of rubrics

Introduction

Working in a journal publishing house, I witnessed many times attempts to create a good rubricator. Most attempts were reduced either to dividing one big rubric into several small ones, or, conversely, to merging several small rubrics into one large one. All attempts to create an ideal rubricator turned into finding a compromise between a complex and very complex rubricator.
I would also like to note that all the rubricators that I have seen were organized in the form of a classic tree with a nesting depth of 2-3 levels. And there were no attempts to organize a rubricator in a different way (We are talking only about printed rubrics).
As a result, I have accumulated a list of questions that have to be addressed by any compiler of the rubricator.

How many rubrics do you need?

Crushing a large section into several small ones always seems like a very simple and logical way to create a rubricator.



However, crushing is always a compromise. On the one hand, there is a simple and clear section, in which there is a lot of information, on the other hand, there are many small sections. The only difference is where exactly a person gets lost - in the elements of one huge section, or in a huge number of small sections (For example, on the site www.ashmanov.com/tech/semantic a universal rubricator for 2500 headings is promoted!).
')

Uneven nesting

Classic “paper” rubricators have a small nesting and a relatively small number of headings.
This allows you to reduce the number of simultaneously visible categories (visible at the current level) and evenly distribute content. However, in rubricators built according to this classical scheme, there is a danger of appearing at the root of the rubricator both final headings, the further division of which is impossible, and headings that are first divided into just several sections, which in turn are divided again and again, t. e. very narrow, but deeply nested headings appear.



The example shows that at the same level the rubricator exists as sections that can be finite “10. Fireplaces. Furnaces ”as well as sections that will necessarily require further clarification.

Other

In the print editions, the section “Others” is always present. The name, which would be logical to read as "Try to guess what is located here."
Interestingly, does anyone even look into it?

How to call a generalizing "rubric fork"?

For many rubrics, tag names are created as names. In the previous illustration, these are “baths”, “saunas”, “pools”. That is, when a separate section seems to be needed for something related to water and bathing, but what should we call it?

Use of highly specialized terms

Sometimes attempts to find this most common name lead to the use of highly specialized terms. Try to guess where in the previous figure is the plumbing section? Not found? And he is not there. But there is a generalized section: "Equipment for engineering systems." Of course, experts will guess that this is probably about plumbing, but are all, who will read the names of the rubricator being created, specialists?

How to describe the dual meaning of some sections?

If everything is good with goods that have one very clear purpose, and it’s quite simple to choose a heading, then with services, especially if the services section is carefully created in the heading (the first section in the previous figure), a real nightmare begins.
Where, for example, place an ad "glue wallpaper"? in the services section or in the wallpaper section? In each of these sections, the ad will be out of place.

One more example:



It seems to be all right, but why is the repair and service of passenger cars located in the car-care center and services section, and not in the passenger cars section?

It's simple. If you create a service section in the “cars” section, then you will have to create the same section for trucks. Why not? But why: the section "cars" is not the final section. And this means that all of these subsections will have to be created for each of the subsections. And this will lead to the fact that there will be a lot of partitions. And for a regular print edition, this is unacceptable not only because of the large volume of the printed rubricator, but also because of the difficulty of choosing a section for publication.

Imagine that you want to rent a domestic car for use as a taxi. And here you have before your eyes the sections “domestic”, “rent” (and it is not known what), “rental”, and “taxi services”. Eyes scatter, well, do not submit the same ad in all sections. As a result, you turn around and leave to submit an ad to another magazine, where there are only two sections - “buy” and “sell” ...

Duality according to the principle of applicability. "Parquet glue or parquet glue?"

In addition to the uncertainty on the part of “goods-services”, there is also uncertainty on the part of the goods themselves and their purpose. For example: parquet glue is unambiguously the “glue” section and just as unambiguously the “parquet” section. But definitely - this is one section, not two! So it turns out an insoluble contradiction.

Whether to make hints to the category in the form of a list of keywords?

And what if for each heading to create your own set of keywords? And then it is possible to unambiguously determine which rubric suits us best. For example, for a two-level rubricator, it might look like this:



A good option. Please note that section 1 has no keywords. It is understood that the section describes the included headings. Look at the name and keywords under category 1.2. There are also clearly visible difficulties with the choice of the name and the formulation of keywords describing the rubric.

And why not add a description to the main sections? And who said that it is impossible? Sure you may. But if we take into account that the section is the sum of all subsections, then this description will be simply huge.
Whether to distinguish end and nodal sections?

Another illustration from the same rubricator:



If we say that a certain document is related to a certain final heading, then we put quite a definite meaning in this statement. But when we attribute a document not to a final, but to a nodal heading, the logical meaning of this assignment is somewhat blurred. In this case, we can both imply that the document has some very generalized meaning and is not subject to further detailing, and that the document can be attributed to all (or almost all) subsections.

Thus, we can formulate several rules for assigning to the top level category:
1. The document refers to the nodal heading in case it cannot be clearly assigned to any of the subheadings.
2. A document refers to a nodal heading if it can be assigned to all or to most of the child headings.