Fourth stage of KM revealed at the KMWorld conference
It was refreshing, to say the least, to be back at the KMWorld conference in person. At the Grand Opening Reception in the Enterprise Solutions Showcase, the high energy level was palpable. The number of sponsors and exhibitors handsomely exceeded that of any previous KMWorld conference, as did the volubility of the conversation. The industry’s perception that KM was a happening field was obvious and exciting. It seemed clearly to reflect much more than just the joy and relief of getting back together again.
A label for this year’s conference could be “Triumph of the Triples”—the triple of “subject, predicate, object” as the basis of RDF (resource description format) and knowledge graphs. This year cemented the dominance of knowledge graph technology in the KM and enterprise search arenas and established it as the fourth stage of KM.
Characterizing the fourth stage of KM is the structuring of data, which is accomplished by creating and deploying the triples of RDF. Stage one of KM was the deployment of the new internet technology to transfer and share information among geographically distributed companies. Stage two was the recognition of human factors issues (it’s no good if they don’t use it) and user involvement. Stage three was the taxonomy stage of data description and retrieval (it’s of little utility if they can’t find it). Stage four is the recognition that, while good description and retrieval capability is essential, it is not enough. To develop question-answering and chatbot systems, it is essential that data and information be structured in a fashion that connects and relates that data with a methodology that is far more capable than what traditional DBMSs and data normalization techniques could deliver. This is where the action is now and where it will be in the near future.
Zombie knowledge
The most intriguing new entrant in this year’s conference was the concept of “zombie knowledge”—how to recognize it and respond to it. Zombie knowledge is what is in your KM system but is no longer useful, pertinent, or even accurate. It is zombie-like because it is built from what was live and valid information, but as information that has not died as it should have, it is now capable of behaving malignantly. And, even if nonmalignant, it is noise and clutter that distracts and impedes the system. Maintaining and providing such information is often counterproductive. How should such information be identified and handled?
Explicit recognition of the zombie notion apparently originatedin a conversation at a SIKM forum last January, catalyzed by a phrase tossed out by Patrick Ready, KMS Lighthouse, and seized upon by Jane Dysart, KMWorld conference chair, and Tim Powell, The Knowledge Agency. The topic was bookended at the KMWorld conference by a workshop led by Powell on Monday and a presentation by Patrick Lambe, Straits Knowledge, on Thursday.