‘Super Search Engine’ Would Simplify the Hunt for Information

Here’s the scene: You’ve just stepped off an airplane in a strange city, and you don’t know where to stay, where to eat or how to find the auditorium where you are expected to deliver a keynote address in half an hour.

You haul out your wireless palm-sized computer and ask for help. But faced with so many questions, the device that was supposed to make your life easier gasps and crashes as it struggles to get through the many traffic jams on the Internet. You end up spending the night in a cardboard box, eating a cold hot dog from a street vendor, and losing your job because you never found that auditorium.

Well, maybe not. But help is on the way.

Researchers at Pennsylvania State University are developing a system that should allow your pint-sized computer to do all of those things by simultaneously searching several databases and integrating the information, even if you didn’t use the right search words. The Mobile Data Access System is designed to make the less powerful tools of wireless communications work more efficiently in the increasingly congested world of global communications. But it also could do wonders for hard-wired terminals, according to Ali Hurson, professor of computer science and engineering at Penn State.

“Think of it as a super search engine,” Hurson said.

The heart of the program centers on a hot-button topic in computer science these days--multiple database systems. Such systems differ from search engines such as Yahoo in that instead of using index terms (or buzzwords, as Hurson calls them) to search for a Web site, they search entire participating databases for keywords.

On the Internet, he said, “you are using those buzzwords to find a proper site which is of interest to you based on the search words you used,” Hurson said. “In our case, we are not looking at the Web sites. We are looking at databases in which the data is well-organized. The structure of the information we have [in a database] is much more complex than the structure of information in a Web site.”

A database could also be a Web site, he added, “but the Web site is a very simplified version of the database.”

Hurson uses this example. Suppose you are interested in sending your child to Penn State or Purdue to study computer engineering. You can use a traditional search engine to find the institution and then search that site for the computer engineering department. But what if you don’t know which school you want to send the kid to, and you would like a survey of lots of schools that offer degrees in that field? Multi-database processing would allow access to databases from all participating schools.

“Computer engineering at Penn State is one data source,” Hurson said. “I made my database without knowing anything about Purdue, and Purdue made their database without knowing anything about us. But you want to get some information about us, Purdue and anybody else who is talking about computer engineering programs.

“So we want to allow these different information sources, which have been created in isolation, autonomous from each other, to be linked together for any party who wanted all this information.”

The system is far from operational, however. Integrating different databases is a challenge, since many use different query, or transaction, languages. Scientists at the University of Bergen in Norway have been working on that problem for years.

But by using a multi-database system approach, a computer could search all available data systems for computer science, and integrate the results in a document that would spell out which schools are offering which kinds of programs.

It could even search for keywords that are similar in meaning to the one you chose. “Suppose I’m looking for the word ‘salary,’ ” Hurson said, and the Internet search engine you are using doesn’t have that particular word on its table. It won’t direct you to a Web site, because it didn’t find one with that buzzword.

But “wage” means essentially the same thing, he added, so the Penn State system would also look for any database with the word “wage.”

“So we have some degree of intelligence in our search engine, based on the semantic response to your request,” Hurson said.

That sort of search is not the kind of thing you’re likely to try with your Palm Pilot, but finding the best hotel in a given city is. Today, if you want to know where to stay, where to eat and what the weather is going to be like, you will have to log on to at least three Web sites.

But a single query to a multi-database system could supply the same information in far less time by simultaneously accessing all three sites, he said. Just type in “Los Angeles,” “hotels” and “weather,” and the gizmo will supply information on all three.

*

For the near future, multi-databasing, known among nerds as “heterogeneous distributed databases,” lends itself especially well to narrowly focused areas such as travel. You don’t need to search the entire World Wide Web for hotels if you know where you are going to spend the weekend. A few integrated databases should do the job.

Ultimately, the Web may look quite different as more and more databases become integrated.

*

Lee Dye can be reached at leedye@ptialaska.net.