Future systems will cull hidden knowledge from seas of data

   ADVERTISEMENT
  

Free IT resource

Hear how top CIOs turn change into a competitive advantage.

Sponsored by HP

Free IT resource

Try Sun servers, workstations and storage products free for 60-days.

Sponsored by Sun Microsystems

RELATED LINKS
»  Google delivers ad-supported video clips via AdSense
»  3Com waiting for details of Bain/Huawei acquisition bid
»  Indian outsourcers' U.S. shopping spree
»  Business RSS feed 

IDG ENTERPRISE NETWORK
• The Broader the E-Biz, the Bigger the Lawsuit  (CIO)
• Time To Change  (CIO)

TOP NEWS 


IT SOLUTION SEARCH

KNOWLEDGE management comes naturally to humans. A 5-year-old with a bag of marbles can divide them up according to size, color, smoothness, and other familiar qualities. A child can even add more categories to the classification scheme as new qualities appear.

Unfortunately, when it comes to information classification, the traditional computer system can't hold a candle to the average 5-year-old. As a result, today's knowledge management systems don't optimally use their potentially lucrative seas of customer, transactional, and other business-related information.

Because traditional computer systems must be programmed with the exact series of steps required to perform a given task, their output is necessarily prescripted, and they are essentially incapable of doing what's really needed: recognizing new patterns in the volumes of data, adapting to new concepts and knowledge as they arise, and assessing the true importance of given pieces of information, not to mention exposing untapped sources of potential revenue and savings. In other words, today's knowledge management tools can deliver only the information that users know how to ask for.

So why can't computers think more like humans? Researchers and software developers are making headway in this direction, particularly with artificial neural networks, a form of artificial intelligence modeled on the human brain. These highly interconnected networks of processing elements can be programmed to adjust the connection strengths, or weights, between processors based on the data they're exposed to. Instead of following a set of instructions to solve a problem, artificial neural networks learn by example.

Once an artificial neural network is trained with examples taken from a carefully selected set of data, it can discover classes, patterns, and relationships in new data sets of the same type. Artificial neural networks are capable of massive parallel processing and aren't confused by noisy or incomplete data, so they're well-suited to churning through large volumes of unstructured information.

Artificial neural networks have been used for all sorts of classification, forecasting, and modeling applications. They have formed the basis of handwriting-, speech-, and image-recognition systems; applications that predict future stock prices, sales demand, and credit risk; and models for optimizing cargo routing, supply chains, and other business processes.

The real payoff in the knowledge management arena is still three to four years away. As neural networking techniques evolve, knowledge management solutions will become not only more adept at distilling and classifying vital information, but will also extract hidden trends and relationships from huge volumes of data, requiring knowledge workers to sift through less information and giving them more relevant information for making decisions.