By Bit Rambusch, vice president, eServices and knowledge management, Dell Technologies
I sat down recently with Seth Earley, sought-after guest speaker and author of The AI-Powered Enterprise to talk about the power of data classification. Seth shared some resonant insights and advice, including the role of taxonomy in structuring data assets correctly: an often-neglected practice (to a company’s detriment).
By way of a definition, taxonomy is the classification of data into categories and sub-categories that can be recognized by computers and help a forward-looking organization create a unified view of information. Essentially, it organizes companies’ data. Using a taxonomy structure, data is tagged to capture the relationship of the data to the parent data. This context makes the data discoverable, and therefore useful.
In Dell Technologies’ recent eBook, Beating the Data Paradox: No Mission is Impossible, we count taxonomists as key players in an enterprise’s in-house data “A-Team.”
This is some of what Earley has to say on the subject.
Written summary
Taxonomies frequently order data into a hierarchy of categories for analysis.
At the very top of the hierarchy is the primary taxonomy. Earley refers to this primary taxonomy as ‘is-ness,’ which gives us a high-level description of what the object is rather than what it is used for. In Earley’s example in the video above, the ‘is-ness’ could be, for example, a legal contract.
However, taxonomy isn’t just used for classification—it’s also used for navigation and to help people find what they need. Navigation taxonomy works alongside classification taxonomy as a set of control terms that enable users to search for things or results in a given category.
Continuing his explanation, Earley refers to the control terms in navigation taxonomy as ‘about-ness.’ If your organization has 1,000 legal contracts, how does it tell each of them apart? The answer is by organizing these contracts into a series of sub-categories that shed more light on individual segments, such as customer name, sector or contract type.
Unfortunately, when employees start building ‘is-ness’ and ‘about-ness’ across the enterprise, they often create multiple taxonomies. The result is many taxonomies rather than one—and a set of organizing principles that should help us produce a more effective search process but can, counterintuitively, end up working against it.
In the video, Earley suggests a complementary approach leveraging ontologies, which achieve higher levels of sophistication by providing richer information and showcasing the relationships between entities. If we return to our legal contracts example, that might include clients who have multiple contracts with a range of departments or those who work across a range of sectors.
He paints a picture of ontology as the “knowledge scaffolding” of the organization. It includes all the big buckets of information we use in a simple taxonomy, but also the organizing principles, the sub-categories used to describe content and all the relationships that link this information together.
Even more importantly, once you have an ontology in place, you can start to think about how you might use knowledge graphs. In a data-driven world where companies are starting to automate strategic decisions based on the volume of customer information they collect, knowledge graphs provide the critical connective framework for extrapolating insights from enterprise data.
When knowledge graphs are built on top of ontologies, businesses can navigate information effectively and retrieve valuable information and insights sooner.
As Earley explains, by integrating taxonomy, ontology and knowledge graph technologies, organizations can draw far more value and perhaps even new revenue-generating opportunities from their data.
And there’s more…
As fascinating as it was, our conversation wasn’t confined to demystifying data classification and organization. Earley also offered helpful advice that any enterprise looking to employ taxonomies and ontologies can use:
- If you’re going to have an avatar, you’ll need to create ‘handles’ so that the avatar can receive information in context.
- When it comes to digital body language, think very carefully about the signals you want to send out.
- Senior executives need to speed up what he calls the “information metabolism” of the organization to remove friction between people and departments.
- In this modern age of AI, new jobs will be created, but humans will still play a crucial interpretative role because of the brain’s ability to recognize subtlety and nuance and to draw inferences from patterns or from stand-alone data points.
You can watch the full, unabridged episode here. I trust it will advance your understanding of what it takes to become a data-driven business and help facilitate your organization’s own digital transformation.