Classifying Merchandise

There are three related yet often confused terms that creep into conversations about retail. I started a conversation on that topic over at the ARTS Linked-in discussion board that has some interesting comments. The three terms are (with definitions from Wikipedia):

Merch Hierarchy.gif

Hierarchy "is an arrangement of items (objects, names, values, categories, etc.) in which the items are represented as being "above," "below," or "at the same level as" one another and with only one "neighbor" above and below each level."

Taxonomy "is the practice and science of classification. Typically this is organized by supertype-subtype relationships, also called generalization-specialization relationships, or less formally, parent-child relationships."

Ontology "is a formal representation of a set of concepts within a domain and the relationships between those concepts. It is used to reason about the properties of that domain, and may be used to define the domain."

Did I loose you on that last one? Ignore it for now. When we classify merchandise, do we use a hierarchy or taxonomy? The generally accepted term in the retail industry is "merchandise hierarchy" but I'm not so sure its accurate.

The classic example of a hierarchy is that of the organization. The lieutenant reports to the captain who reports to the major in the Army hierarchy. For a taxonomy, the classic example is the classification of animals. The dog is a canine, carnivore, mammal (sparing you lots of Greek and Latin words).

So which term fits Electronics --> Televisions --> Plasma --> Panasonic TH-50PX60U best? Looks like a taxonomy to me because each successive layer gets more detailed. I'm amazed at the lack of standards in this area. Supplier/retailer data pools should be much simpler than they are.

Which brings me to my final rant. Until we, as an industry, straighten this out we'll never achieve the machine-to-machine communications we strive for in the Semantic Web. This is where the term "ontology" becomes relevant, as it provides the discipline to define a constant view of the retail domain that moves us from information to answers.


As the web & mobile become even more central to shopping and commerce, I expect solving this problem will become even more critical.

I've had to reconcile and work around various "merchandise hierarchies" (or "category hierarchies," as we knew them) over the last few years, and it has occurred to me that the standards-authoring organizations (e.g. the ARTS) might do well to hire an information architect to simplify things.

It will be nice when our machines can all speak to one another, but the benefits to the shopper will be even more powerful.

Posted by Keith Anderson on October 26, 2009 at 05:36 AM CDT #

Actually, there are some standards--some old, some emerging. It's just that retailers always have tended to group products by buyer skills or how they are displayed rather than any science. And for some categories (e. g. books) the existing available hierarchies (e. g. BICS, Dewey) aren't "retail ready"

As a start point:

The obvious examples are GATS Harmonized Tarriff Schedule (which really isn't that harmonized around the world, alas), or the standard that seems to be gaining traction: UNSPSC "United Nations Standard Products and Services Code", which is maintained by GS1 US.

Most major tax engines support this UNSPSC, as does Oracle eBusiness suite.

Also, NATO runs a classification based on a detailed "fit-form-function" product description (SKU or NSN number per unique fit-form-function, as described in detailed "Item Identification Guides"). Good for things like hardware and tools...


Posted by Miles Thomas on November 08, 2009 at 11:21 PM CST #

Nice note and I concur with the point about the need for ontologies.

Elaborating on the comment by Miles, Oracle Spend Classification from EBS Supply Chain (OSC), does work extensively with taxonomies and provides
intelligent, automated classification capabilities.

Out of the box it handles a few standard taxonomies including UNSPSC and EBS's own. It can classifiy txnl. spend data against custom taxonomies.
It also is integrated with Spend Analytics (OBIEE).

OSC has a hierarchical classification engine that leverages oracle datamining under the covers and performs intelligent hierarchical classification.

The engine is driven by user specified attributes and learns from new patterns that appear in the data.

One of the lesser known facts is that this classification engine is not restricted to doing spend.

If required, it can be repurposed to do other kinds of classifications (e.g. Supplier classification, Customer classification etc.).

keshava rangarajan

Posted by Anonymous on November 09, 2009 at 01:22 AM CST #

Solving this problem is critical these days. Having no classification can be disastrous for some companies.

Posted by Dave on March 29, 2010 at 03:39 PM CDT #

Post a Comment:
  • HTML Syntax: NOT allowed

David Dorf, Sr Director Technology Strategy for Oracle Retail, shares news and ideas about the retail industry with a focus on innovation and emerging technologies.

Industry Connect

Stay Connected


« November 2015