Search researchers say the hardest job of a search engine is not in the finding of data . It is in the delivery of highly customized results that have been pared down based on the user's context. Still, while search engines like Google, Yahoo and Ask Jeeves inch closer to returning word-based context-oriented results, image-based tools are in development that appear to have promise.

Good Results for a Computer

At Princeton University's Shape Retrieval and Analysis Group, professor Thomas Funkhouser helped to develop a 3D model search engine. It has been on the Web since 2001. While it may not be practical for the everyday Internet user, the technology may have near-term applications in such areas as engineering, inventory, mapmaking and design.

"It currently indexes more than 30,000 models," Funkhouser told NewsFactor. The project is designed to study 3D models and the retrieval of shapes. The search engine works by drawing a shape with a mouse and hitting the search button. Users also can add text to the search to narrow the results.

Funkhouser and the team are interested in developing an effective query interface and building shapes that return good results. "We use statistical methods to index and match 3D models," he said. "The computer's processing does not match what humans do with their minds. However, the results can be quite good when matching distinctive 3D shapes," he said.

Baby Steps

Search is essentially the quest for artificial intelligence, say researchers working on commercial search engines. It certainly is a sexy idea -- becoming the parent of the first thinking machine.

But it could take much longer than one might imagine. The human mind is able to make rapid associations and deliver specific results with only a few pieces of information. In the realm of words, a computer obviously can store massive amounts of data, all equally accessible in linear arrangements.

But machines fall behind, compared to a person's brain, when it comes to assigning context to data. That is why search is still taking baby steps toward offering efficient results that match the flexibility of the individual user.

In the world of images, computers are hopelessly lost. Considering that an ordinary individual can identify a briefly seen human face from among tens of thousands, the performance of the human mind is staggering. But it is just not speed and accuracy that challenges the developers of visual-based search. (continued...)

1  |  2  |  Next Page >