Authored by Peter Dahl, Senior Consultant, Utopia, Inc.
It's funny how looking at different parts of a compound word gives different connotations. In our taxonomy world, parametric has nothing to do with the metric measurement system ...it's all about parameters. Parameters are the "old" version of our current terms attributes or characteristics. However, parametric also implies that there is order to the characteristics.
Why would we care about order? Ok, that's a tongue-in-cheek question; order is invariably a good thing. Seriously though, nowadays we seldom manage to run any data system - whether it is a mom-and-pop shop with a simple database of retail products, or a full-fledged ERP - without considering the value of an excellent taxonomy. And, until someone creates and donates a worthy full-fledged taxonomy to the public domain, it's usually developed in-house.
A good taxonomy will include all the elements that allow the advantages of database tools - a noun and modifier (or class), attributes (or characteristics), a short description and a longer, fuller description. Ideally, the attributes are ordered. Why ideally? Well, let's think back to my parametric term. The implied order in the list of attributes allows the use of the noun and modifier (class) and the attributes (characteristics) to automatically generate the full description.The order dictates, then, that the description will contain the relevant information about the object from most to least relevant. Why? Well, again, we can use database tools to make comparisons, and allow the use of fuzzy logic in searches. If the data in a text block flows from most to least relevant, the most relevant attributes can be matches, while the least relevant can be fuzzy matches ... which allows one to find something "close enough" in an emergency. Parametric data also allows for the most effective reviews for potential duplications.
The critical thing, of course, is the order ... and it is quite difficult to arrive at a perfect order. Five taxonomy experts will likely have five opinions on the order of any specific item. To confirm that, check out product catalogs for five different bearing companies and you'll very likely find that none agree on all the attributes, or on the order of relevance. My rule of thumb for dictating the order of attributes is that each successive attribute should narrow the set of possible matches the most. For example, the noun bearing will contain all possible bearings. The class roller bearing is an immediate huge reduction in the bearing set. One then needs to look at OD, ID, width, material of construction and other such relevant attributes to determine which will reduce the set of potentials most.
While one can only dictate the ordered attribute set and the resulting parametric description for one's own business, a well defined and controlled taxonomy will allow the use of standardized tools to reduce the variability of data and/or goods as well and helping ensure the data/items can easily be found and harder to inadvertently duplicate.
Why would we care about order? Ok, that's a tongue-in-cheek question; order is invariably a good thing. Seriously though, nowadays we seldom manage to run any data system - whether it is a mom-and-pop shop with a simple database of retail products, or a full-fledged ERP - without considering the value of an excellent taxonomy. And, until someone creates and donates a worthy full-fledged taxonomy to the public domain, it's usually developed in-house.
A good taxonomy will include all the elements that allow the advantages of database tools - a noun and modifier (or class), attributes (or characteristics), a short description and a longer, fuller description. Ideally, the attributes are ordered. Why ideally? Well, let's think back to my parametric term. The implied order in the list of attributes allows the use of the noun and modifier (class) and the attributes (characteristics) to automatically generate the full description.The order dictates, then, that the description will contain the relevant information about the object from most to least relevant. Why? Well, again, we can use database tools to make comparisons, and allow the use of fuzzy logic in searches. If the data in a text block flows from most to least relevant, the most relevant attributes can be matches, while the least relevant can be fuzzy matches ... which allows one to find something "close enough" in an emergency. Parametric data also allows for the most effective reviews for potential duplications.
The critical thing, of course, is the order ... and it is quite difficult to arrive at a perfect order. Five taxonomy experts will likely have five opinions on the order of any specific item. To confirm that, check out product catalogs for five different bearing companies and you'll very likely find that none agree on all the attributes, or on the order of relevance. My rule of thumb for dictating the order of attributes is that each successive attribute should narrow the set of possible matches the most. For example, the noun bearing will contain all possible bearings. The class roller bearing is an immediate huge reduction in the bearing set. One then needs to look at OD, ID, width, material of construction and other such relevant attributes to determine which will reduce the set of potentials most.
While one can only dictate the ordered attribute set and the resulting parametric description for one's own business, a well defined and controlled taxonomy will allow the use of standardized tools to reduce the variability of data and/or goods as well and helping ensure the data/items can easily be found and harder to inadvertently duplicate.
