杏吧原创

USING search engines to compile a list 鈥� like the top 50 greatest blues guitarists by record sales, say 鈥� involves a lot of drudge work because you have to visit many web pages to gather the data you need. But the next step in search engine technology could make creating such lists possible with a single mouse click.

KnowItAll, a search engine under development at the University of Washington, Seattle, trawls the web for data and then collates it in the form of a list. The approach is unique, says its developer, Oren Etzioni, because it generates information that probably doesn鈥檛 exist on any single web page. The US Department of Defense鈥檚 research arm, DARPA, and Google, are so impressed that they are providing funding for the project.

Etzioni鈥檚 ultimate aim is to have KnowItAll answer questions such as 鈥渓ist all British scientists born before 1900鈥�. The software cannot do that yet, because it lacks a module that can understand 鈥渘atural-language鈥� questions of this type. That will come later, he says. What it can do, however, is take a phrase like 鈥渓ist scientists鈥� and return with a list that it believes with a high degree of confidence are (or were) scientists.

For any input noun 鈥� 鈥渟cientists鈥�, 鈥済uitarists鈥�, 鈥済ardeners鈥� or 鈥渁ctors鈥�, say 鈥� KnowItAll tries to find sentences on websites that contain that noun and looks for words that often appear after it. In this way it might find the phrases 鈥渟cientists such as鈥� and 鈥渟cientists including鈥�. It then feeds these to 12 search engines and extracts the words that tend to follow, which are often scientists鈥� names.

But c1ertain phrases like 鈥渟cientists such as botanists鈥� also fulfil the search criteria. The software can work out that 鈥渂otanists鈥� is not a name, and it can use this to inject 鈥渂otanists such as鈥� into the engines to obtain an even fuller list of scientists鈥� names.

KnowItAll then returns a long list of scientists鈥� names 鈥� each one accompanied by its percentage probability of being correct, as measured by frequency of occurrence of the names on websites. Users will be able to choose the level of confidence they want in the data.

KnowItAll is also able to find words that often occur close to the search term. In the case of 鈥渟cientists鈥� these might be words like 鈥淒NA鈥� and 鈥渜uantum鈥�. It uses them to refine the probability that a person is indeed a scientist.