Bottom-Up Knowledge Capture

[This article originally appeared in the Millenium Issue of Knowledge Inc., The Executive Report on Knowledge, Technology and Performance.]

Knowledge management has been, up to now, largely a top-down enterprise. Driven by a concern that corporate knowledge repositories would quickly fill up with inaccurate, useless junk without rigid quality review, organizations have created small priesthoods of knowledge administrators responsible for virtually all authoring. Unfortunately, the result often has been massive bottlenecks as content generated in this centralized way sits for weeks or months awaiting review. By the time knowledge reaches its intended users, much of it has aged to the point of irrelevance.

Top-down knowledge management has had limited success. KM will begin to show significant ROIs when the process is inverted. Centralized knowledge administration clearly produces higher-value knowledge — but centralized authoring retards growth. In the coming decade, the hard dollar value of knowledge will be recognized, and everyone — not just a small elite — will be responsible for generating the raw materials for corporate KM.

Bottom-up knowledge generation will have significant impacts on the way work, and workers, are perceived by corporations. Management will have to develop new incentives for knowledge workers to contribute high-quality content. For more traditional firms now adopting KM practices, decentralization of knowledge generation will be difficult, as it is antithetical to some ingrained management principles and habits.

The bottom-up knowledge capture trend will have a direct impact on technology. The technologies developed to capture and publish knowledge all have had to compromise one way or another between simplicity and specificity. Relatively unsophisticated techniques like text search are easy and cheap to apply; gather up a collection of documents and point a search engine at them, and in a simplistic way you’ve created a knowledge base — but the results are often not very specific.

Artificial intelligence techniques, by contrast, impose a great deal of structure on knowledge content, making it much easier to find the specific knowledge needed. Structure not only makes knowledge more useful, but also makes it easier to evaluate for accuracy and utility. But the knowledge is expensive to generate and maintain, and these techniques do not lend themselves to generalist authoring.

More advanced approaches incorporating pattern recognition and associative techniques such as neural networks, along with increasingly sophisticated textual and semantic analysis approaches, are enabling KM systems to capture knowledge in more intuitive ways. These technologies support decentralized authoring, not only by simplifying the way structure is imposed on the content but by allowing “authoring” to be a more innocuous, almost passive element of another task, such as solving customers’ problems or answering their questions about products. However, these newer technologies tend to be more expensive to license and implement.

The bottom-up knowledge trend is good news, then, for the technologies at the opposite ends of the spectrum — low-end search tools, which can provide cheap solutions requiring virtually no knowledge engineering, and high-end, high-priced solutions that allow effective knowledge capture to be seamlessly integrated into high-priority corporate workflows. The trend clearly is bad news for vendors relying on older, more labor-intensive structured knowledge representation approaches.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s