Magazine article CRM Magazine

New Paradigms Bring Value to Knowledge Management: The Focus Shifts from Managing Knowledge to Empowering People to Respond, as Seen in Online Communities

Magazine article CRM Magazine

New Paradigms Bring Value to Knowledge Management: The Focus Shifts from Managing Knowledge to Empowering People to Respond, as Seen in Online Communities

Article excerpt

ON JULY 29, Oracle acquired InQuira--the last independent knowledge management vendor in the CRM world.

In the way we talk about it, knowledge is useless. Seriously, there is no value whatsoever to the knowledge we produce and manage via knowledge management. While this statement is bound to anger most of my vendor clients and confuse end users, please hear me out.

Without going too far in time, we can safely say we started this knowledge management trend 25 to 30 years ago when we discovered we were no longer "industrial era" workers but, rather, knowledge workers. The theory was that if we were valued for our ability to create, store, and reuse knowledge, we could build systems that would allow those around us to leverage that knowledge. Fast-forward 10 to 15 years, and the automation of this knowledge via self-service systems became one of the first things we could envision the commercial Web doing well.


Can you imagine, we said, if any client could find what he needed without the help of an operator (back then, we were still mostly phone-based--same as today, actually)? The cost of a self-service transaction was quoted by vendors to be as low as 1 cent. One cent! There was no way you could beat that. We set out to solve the main problem we thought we had: how and where to store the data, and how to find and retrieve it.

Hardly simple problems. In fact, we had been battling those questions for the better part of two decades in helpdesk applications, call center packages, and even help systems embedded in software applications. We had mastered the art of indexing content and doing Boolean searches by keywords, by phrases, or--the revolution back then--by natural language. The explosion of self-service solutions brought a wave of investments into natural language processing (NLP) tools and solutions.

This influx of cash helped advance NLP, with some of that money devoted to figuring out the problem of how to use speech as an entry point for NLP-based systems. With new models and techniques for NLP and speech recognition, we were able to advance from 40 percent to 45 percent out-of-the-box comprehension without training to between 50 percent and 60 percent. Even better, with appropriate training, we could increase the understanding to almost 80 percent in certain cases.

However, this enhanced recognition did little for self-service. Adoption was great at first, as people rushed to deploy solutions seeking to cash in on those low costs per transaction. That great reception was tempered almost immediately, however, as reports emerged that self-service solutions were unable to find the right information. New models for navigation and searching were used and NLP continued its progress, but the right answer remained elusive.

We realized that the problem was not the search, but what we were searching. The stored knowledge that the NLP queries were retrieving had two major problems:

It was obsolete; I am fond of saying that data and knowledge have depreciation rates of 50 percent per minute. This means that after two minutes, the value approaches nil. That may be tongue-and-cheek, but in reality stored knowledge does not fare much better. Organizations that deployed self-service solutions quickly found that their knowledge could not answer the questions before it had to be changed.

It was inaccurate. Even when knowledge is created and stored within a time frame that would allow it to be used before it becomes obsolete, the chances that it is accurate and responds to the query are small. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.