Emerging technology: Agents, algorithms, et al. -- ADTmag

Emerging technology: Agents, algorithms, et al.

November 29, 2002

You are right: Agents, algorithms and heuristic avatars are less than new. When we consider such components, the focus is on the "emerging" in emerging technology. Some technologies are ever emerging, but they still bear watching.

»Intro
»Grid computing
»Rich clients
»MDA
»Wireless computing
»Agents, et al
»Burke's laws

For a number of years, Artificial Intelligence (AI) was the premier emerging technology to watch.

Though less touted of late, AI research has never stopped, and a fresh generation of technology workers is ready to consider it anew. At the heart of every computer system are programmatic algorithms, and there is every reason to think that writers will make breakthroughs in this black art.

A real enabler is the industry's standardization on Java middleware and platforms, which has allowed a more vibrant component market to grow. There was a time when an inventor with an algorithm needed to program a system from the ground up. Lately we have seen rules engines and the like that plug into Java app servers bought off the shelf.

To look at a single example: Advanced Software Inc., in business since 1986, has a patent on an apparatus and method for comparing data groups. The work is embodied in its DocumComp Comparison engine. And this is in technology that is, in turn, at the heart of one of those useful doodads most of us have yet to uncover in Microsoft Word. This fall, Percussion Software said it would integrate DocuComp into its Rhythmyx Content Manager. Advanced Software also announced that the San Francisco-based non-profit Internet Archive licensed DocuComp to compare Web pages on its Wayback Machine Web site archive (http://www.archive.org).

What is inside DocuComp? Said Advanced Software CTO Chuck Grasso: "It is a multitasking algorithm that uses iterative refinement" to compare documents at ever-finer levels of granularity. The engine interfaces with Java, said Grasso, who noted that the system can compare about 20,000 pages per minute.

Agents and algorithms coda: Rules-based systems have enjoyed a few incarnations, but being the "keeper of the rules" has never been a sound job path. That is an inhibitor. Is a rule an improvement on a present solution? Usually, the good ones are successful only as encapsulations of corporate wisdom that have already been acquired, but deft implementation of such formulas can save money for companies. The problems good algorithms solve are the most common problems; thus, they have far-reaching effect. Savvy ISVs that can spot talent are the best at exploiting such solutions, and mature platforms are now ready to help in this endeavor. But Web services could change that: Think of how Google or Amazon.com changed your workday; then think about connecting up via public SOAP APIs.

Prev | Next