Embedded Information Systems Will Make Cars Talk, Anticipate Shoppers' Needs And Track Our Every Move <!-- Newstype: Latest_news

BUFFALO, N.Y. -- The next area of explosive growth in information technology will be in the field of embedded information systems that more or less read your mind.

That is the prediction of Joseph D. Woelfel, professor of communication at the University at Buffalo, who nearly 20 years ago developed CATpac (CATegory PACkage), a unique embedded system of information retrieval and analysis now used by researchers world-wide.

Embedded information systems are able to read and recognize individual neural patterns and from them predict behavior. The nature of these systems offers a glimpse into the complexity of the new electronic age, according to Woelfel. They point to the speed with which vast technological changes inform our lives in ways of which most of us are hardly aware.

Whether we know what they are or not, within three or four years, Woelfel says, these embedded systems will provoke enormous changes. He notes the possibility of:

• Cars that will talk to us and accept verbal commands, predict our destinations and get us from one place to another by telling us where to turn and warning of detours and accidents ahead. They'll remember that you often forget to turn on your headlights, for instance; read the level of outdoor light, and query you.

• Household appliances that not only will recognize and remember the way we like our coffee brewed or our bread toasted and provide both automatically, but will tell us when they need new parts.

• Commercial applications that will recognize the way we usually shop, anticipate our needs, suggest items that might interest us, advise us of pertinent sales or ask us before we leave the store if we forgot to buy shampoo.

• An array of satellite networks feeding data into computer networks that not only will individually and collectively see and track us, but know what we are likely to do or where we are likely to go next.

"All of this raises new and difficult ethical, legal and philosophical questions that must be examined carefully and with considerable knowledge of the rapidly changing field of communications technology," Woelfel notes.

"This is a very difficult job because such changes will not only be endemic, but will provide information to strangers about many different areas of our lives," he says. "On top of that, much of the programming is 'hidden.' The average person may become aware of new applications, but won't know where they came from, who imposed them and why or how else they may be applied."

The embedded systems of which Woelfel speaks are intelligent systems that are inside other systems and are not themselves new. They see what information is requested and, based on that information, make additional suggestions to the user. For example, the World Wide Web is an intelligent system and a search engine is an intelligent system embedded in the Web.

There are two ways these systems work, according to Woelfel.

The first, and weaker method, he said, is "collaborative filtering." This is the one with which computer users are most familiar. It collects information passively by recording what you select on your computer, or actively by "asking" you what you want or where you want to go next. It then measures and compares the collected data and finds other customers who searched the sites like those you've searched or bought what you are buying. Then they suggest to you the choices made by these other users.

"This method is popular commercially," Woelfel says, "but it will only come up with the most obvious choices -- the same authors' names, Web sites or broad interest areas, for instance."

The second, newer and much better method by which embedded systems can work allows them to recognize how we think and make predictions based on our neural patterns.

"We now know enough about human cognitive functions to simulate them in a computer," Woelfel says. "Using a method of simulating neural patterns, the computer can learn an individual's behavior pattern -- what kinds of information he seeks, what she buys, what channels the user watches -- and can predict with considerable accuracy what that user will do next.

"It isn't anything nearly so complex as the human brain," he adds. "These are rather simple artificial neural networks, but they work very, very well. By recognizing part of the pattern, they correctly predict the whole."

In addition to helping us by organizing, prodding and reminding, some the most popular commercial applications, like the one developed by Woelfel, make new and highly accurate forms of market research possible.

Although the communications revolution frequently is compared in importance to the Neolithic Revolution, the Industrial Revolution and the vast changes that followed migration from rural to urban areas, Woelfel goes further.

"I think this revolution is bigger and even more important than the others," he says. "It is changing our lives in unprecedented ways, and most important, with unprecedented speed. We all know that we barely have time to become familiar with one new technology when a new, improved, faster, better, more complex system arrives on the scene with its benefits and problems. And this is happening in all fields at once, of course, provoking more and more change on every level."

While admitting the amazing number of opportunities being created daily in the jacked-in world of information technology, Woelfel, like most experts in the field, emphasizes that a change this massive provokes social, political and cultural questions that we barely have begun to address.

"You know, if you carry a cell phone," he says, "your service provider knows where you are within 30 inches. Lawyers now subpoena cell-phone records to track the exact location of individuals at a particular time on a particular date.

"The police want access to these data files, too," he adds. "Commercial concerns want access -- it's intrusive, yes, but then again, if you're buried in a snow storm, your location could be pinpointed immediately within a few feet."

In addition to the application of value systems to technological innovation, some of the greatest IT headaches we'll face as we plow through the '90s and into the next millennium, says Woelfel, are the staggering problems presented by the need to retrieve data from old hardware and from discontinued software programs, so that it isn't lost.

"We also have to rewrite existing programs to accommodate new operating systems and software. That is definitely a challenge. I spend a good deal of time updating the CATpac programs (see box) used by thousands of analysts and researchers. It's tedious and boring work, but it requires technological skill. So that's another problem to solve."

Before using new technologies to predict behavior, the user must input raw data into a program. Virtually all embedded systems can read only numerical data, although a vastly greater amount of data is available only in text form. These text data would be virtually useless were it not for Woelfel's copyrighted CATpac program.

Developed by Woelfel in 1980 through private funding, CATpac remains the only embedded system that can produce quantitative analysis of qualitative data. It can identify, retrieve, organize, analyze and compare vast amounts of data from text files, and in seconds map it numerically in a variety of useful ways, including tables, charts and three-dimensional rotating graphs.

Today, CATpac is used all over the world to analyze interview data and predict market trends, political attitudes and find the most useful methods to sell the same product or the same idea to an Indonesian or a French cabby or to school children in Zaire. It is used by advertising agencies, by roughly half of the top 100 market-research corporations world-wide and by virtually all major automobile manufacturers.

The Center for International Forestry Research, headquartered in Bogor, Indonesia, is the premiere international organization devoted to the promotion of the biodiversity, social sustainability and balanced management of the world's forests.

It has installed CATpac programs at more than 1,000 of its research sites on six continents. The programs analyze the text of interviews conducted with large numbers of rainforest dwellers in places like Cameroon or Brazil. They then tell the user which arguments and incentives are most likely to provoke indigenous forest dwellers in a variety of different settings to work actively to preserve their environments.