Misunderstanding what classification is and how detailed it needs to be

Since the launch of nsiklab.lt, a website for the Lithuanian AEC market players to familiarize with the upcoming national construction classification system (NSIK), I’ve been getting emails with various questions. While most are relatively unique, I thought it might be worth writing about the common issues, misunderstanding of concepts. I would summarize the prevailing issues as:

• Misunderstanding of what classification is and how it differs from identification.
• Constrained, narrow thinking at the company/organisation level and not seeing the broader picture of NSIK.

I’ll start with the second point first. The theme I am noticing from the questions is the desire to “pull” NSIK towards the company workflows as to avoid changing anything. This comes mostly in the form of wanting to use classification in very niche and very narrow ways. So narrow in fact, that it defeats the purpose of having a classification system and second, completely ignores the fact that other companies and organizations might have diverse needs, different workflows.

As was the case with the inquiries from the same sector. Trying to promote certain classes, ideas as something that needs to be updated in the NSIK for the benefit of all. Yet, those ideas were not compatible, partially proving my point.

When the team began adapting ISO 81346 in the form of NSIK, one core principle we had, was to generalize and avoid tailoring to just a specific subsector. The issue is not that a few classes don’t “fit” ones needs, but that the entire NSIK framework must be as compatible as possible with existing regulations, national needs, national asset managers, IT systems and AEC market. Not a trivial task, our result is far from perfect, but considering the resources, time allocated and the heavy bureaucracy of the project, I think the result is fair to all market participants.

Back to the first point, to better understand what classification is, I’ll try to explain it in completely non-technical terms.

Imagine that your bedroom dresser is a classification system and drawers (infinitely large) are higher level classes/groups for lower lever classes. Now imagine we have socks as a class. How many drawers do you really need for socks? Usually, one is more than enough. Does one really need separate drawers for various types of socks? If you say yes, I want separate drawers for winter / summer socks, when why not a separate drawer for short / long socks? Maybe even separate drawers for white / black / coloured socks? Even better, why not separate drawers for all of these possible combinations?

It is easy to see that common sense is lost. We don’t need that many separate drawers (classes) for all the possible types of socks. But what would make more sense is to have dividers inside the drawers to enable our socks to be placed separately from each other based on their type. This divider essentially enables identification (ISO 81346 aspects) through the use of the type aspect. Currently, NSIK has some standardized types but also a mechanism, for creating custom types by the users.

Alternatively, we can take another example with trash sorting bins. It does not make sense to have hundreds of bins for all the possible types of trash, so we combine them into certain groups based on some shared properties that we subjectively agree are the key ones.

At some point common sense has to kick in and an imaginary line must be drawn that separates where classification ends, and identification begins. While the desire for very definitive and descriptive data will always be there, we must limit ourselves somewhere. Classifiers are an attempt to reduce the available vocabulary for object descriptions while simultaneously grouping them by certain major features/characteristics. We have various instruments from EIR, LOIN, IDS to allow us to better control what data we need in the project. NSIK is information and it can be adjusted based on what is minimally required for a project.

The designers/contractors are always welcome to use their own custom attributes and assign data used for their internal workflows as they see fit. However, these workflows are often not identical between companies and tailoring NSIK towards them would not make sense or be fair to all AEC market participants.

Summing up, classes are supposed to be quite broad and “open” as they serve the role of large container where numerous terms/objects can be placed in. When working with ISO 81346 based classification systems, one must keep in mind that it is a multi-faceted classification system. Hence an object is not limited to the description of only a single class.

We can separate certain data from the object and classify it separately by using different classification tables. For example, NSIK allows an object’s material to be classified by a separate class, while focusing on the object in terms of functional, technical systems and components. Or even attach other information from a total of 15 existing tables.

PS. For the time being, I am safe keeping the suggestions I received for NSIK. When NSIK becomes fully official, I’ll hand those suggestions over to the official institution managing NSIK and also make a larger post/article with these suggestions and my commentary. If you have any suggestions in terms of NSIK, please do send them my way. I collect all suggestions, regardless of how good or poor they are. In the end, it is not up to a single person to make decisions on what will be incorporated into future NSIK versions. It will be based on consensus.