Organizations still need to codify their tribal knowledge, make it findable for as wide a range of use cases as possible, and implement it in orderly processes that heighten productivity and efficiency. “KM, to me is, whatever your goal is, whether it’s better process and efficiency, that means great checklists, great know-how, great templates, [and] great things that guide [us] to do our jobs better,” remarked Alex Smith, global product management lead for iManage RAVN. “But equally, we need to understand what the context of that is.”
Taxonomies
The templates and best practices Smith alluded to are the manifestations of KM that persist regardless of which technologies support it. Often, the context he referred to is informed by department-wide or enterprise-spanning taxonomies—the hierarchies of terms and definitions of how enterprise knowledge is stratified. When working with documents, contracts, applications, and forms, taxonomies provide “that metadata layer to bring context to it all, regardless of where it lives,” Taliano said.
Such metadata is essential for classifying enterprise content, finding it via search, and, to a lesser extent, enabling organizations to provide natural language queries of that content. Contemporary KM resources couple traditional machine learning models, which crawl through sources to create metadata tags based on taxonomies, with language models. With the latter, “We’re leaning on LLMs and more generic language models to interpret the content and understand the information we’re looking for and extract that from the documents [to] build that [metadata] model, build that graph for us,” Taliano revealed.
Folksonomies
As venerable as taxonomies are for KM, even they are being augmented by the concept of “folksonomy,” a term coined by Thomas Vander Wal in 2004 to describe collaborative, social tagging activities that result in user-generated metadata. The folksonomy approach is regarded as being more inclusive than that of taxonomies and, perhaps, more fluid. “A folksonomy is a way of describing your content and building a graph that isn’t as rigid as having to fit in your defined taxonomy,” Taliano said. “Some of it is user-defined. So, [it’s] systems where you add your own labels, or keywords, or tags. This builds a less rigid structure.” It’s likely not a coincidence that folksonomies have re-emerged in KM, alongside the widespread use of GenAI.