Taxonomies and Ontologies Transforming Knowledge Management
Ever since the inception of knowledge management, there have been varying definitions, connotations, and denotations of what KM means to the enterprise. Nearly all of them have relied on taxonomies and ontologies as the basic foundation for compartmentalizing knowledge to make it readily accessible and useful.
According to Marjorie Hlava, chief science officer and chairman of Access Innovations, “Knowledge management depends on figuring out who has what knowledge, what kind it is, and then making a list of it, organizing it, tagging any papers someone’s written, and those kinds of things.”
More often than not, the hierarchies of terms (and their definitions) that taxonomies furnish are the basis of these activities. They provide the terminology by which enterprise knowledge is codified. At a higher level of abstraction, taxonomies provide the nomenclature to describe the particular schema of what makes data meaningful to an organization. This “worldview” of a particular organization is a conceptual data model, or ontology.
Thus, when assessing the merit of taxonomies and ontologies in transforming KM, it’s important to consider a pair of developments. The first is KM’s long-standing reliance on these constructs. According to David Clarke, Squirro EVP of semantic graph technology, “Taxonomies and ontologies started out in life as a better way to get improved information retrieval. Eighty percent of our clients are using taxonomies that way.” Several of the characteristics Hlava ascribed to KM are aligned with making enterprise knowledge searchable.
The second development impacting the relationship between KM and taxonomies and ontologies involves AI. This indicates a subtle shift that could potentially redouble the value of KM to the enterprise. “Taxonomies and ontologies are a way to capture human knowledge in a machine-readable way,” Clarke commented. “As we enter the age of the flourishing of AI, making enterprise knowledge machine-readable is the key enabler for business process automation, not just information retrieval.”
Taxonomies
If the result of the transformative nature of taxonomies and ontologies is automating core business processes, as Clarke implied, it’s spawned from the ability to make enterprise knowledge searchable. You cannot achieve the former without the latter, making the faculty for search no less transformative—and indebted to taxonomies—for KM. Without structured vocabularies, thesauri, synonym management, and other mechanisms that support taxonomies, organizations would rely on ad-hoc, unsustainable means of organizing their knowledge.
By effectively standardizing the terms by which enterprise knowledge is collected and stratified, taxonomies produce several transformative benefits. “People use them in peer review for sorting articles,” Hlava said. “They use them in sorting conference tracks. They use them in support of call centers. They use them in web navigation and webpage keywording. They use them to tag documents: not just the entire document, but the paragraphs, the figures, and the graphics.”
These applications are based on the structured terminology of taxonomies to describe enterprise knowledge. Most make the underlying content searchable. More granularly, taxonomies supply the words in which people think about knowledge, classify it, and, ultimately, make it available.
The transformative effects of taxonomies on KM are expanded when they’re coupled with ontologies. However, taxonomies need not involve ontologies. “If you’re just trying to organize data for knowledge management, the paths of other people, you don’t need the ontology,” Hlava remarked. “You only need the taxonomy.” In addition to providing schema and a conceptual data model, ontologies furnish an orderly structure for adding additional classifications and attributes to business concepts and entities. It behooves organizations to implement ontologies in standards-based environments—such as RDF and semantic knowledge graphs—in which constructs like OWL are designed for users to customize ontologies with their data.
Companies and Suppliers Mentioned