Search engine makes social calls
New algorithm exploits community structure of the web.
The web has spontaneously organized itself into communities. A new search algorithm that pinpoints these could help surfers find what they want and avoid offensive content.
Page builders can link anywhere. But they dont, Gary Flake, of the NEC Research Institute in Princeton, and his colleagues have found. Instead, pages congregate into social groups that focus most of their attention on each other.
Web directories compiled by hand, such as Yahoo!, recognize this to an extent. Flakes team has automated the process. “We find extremely high-quality sites that Yahoo! and Google dont know about,” he says.
The new search ignores a pages text, looking only at its links. It crawls from a starting page to others it links to, and so on out into the web, picking out islands of expertise in the sea of information.
A test of the algorithm starting at the home pages of biologist Francis Crick, astrophysicist Stephen Hawking, and computer scientist Ronald Rivest yielded groups of sites that are tightly focused on each researchers life, work and field. “The sites are remarkably topically related – the clusters properties are completely intuitive,” says Flake.
“You can extract a lot of meaning from links,” agrees Mike Thelwall, who studies search engines at the University of Wolverhampton, UK. The new approach is, he says, a clever way to find meaningful groups among the effectively infinite number of ways to subdivide linked pages.
The first application of community searching may be to fence off areas of the web such as pornography or hate-speech communities, says Flake. Current content filters are largely text-based; these are easy to dodge and require intensive human management.
Google pioneered the use of links to deduce pages relevance. Its PageRank technology counts a link from site A to site B as a vote for B from A. But it does not take account of all the other sites to which A has links, as NECs new technique does.
Flake does not expect to displace the market leader – “Googles a great search engine,” he says. Rather, he wants to add an extra dimension to searching.
Using link structures could lead to more efficient, customized searching, particularly for scientists, who are careful to link to each others pages. “For academics its going to be a big improvement,” says Thelwall.
Flakes team is now mapping out communities in the web as a whole, without using a starter page. This could detect hitherto unsuspected communities, says computer scientist and network researcher Jon Kleinberg of Cornell University in Ithaca, New York.
“It could bring together people with common interests that may not know of each others existence. You could also catch the early stages of new trends,” says Kleinberg.
- Flake, G. W., Lawrence, S., Giles, C. L. & Coetzee, F. Self-organization and identification of communities. IEEE Computer, 35, 66 – 71, (2002).
All news from this category: Information Technology
Here you can find a summary of innovations in the fields of information and data processing and up-to-date developments on IT equipment and hardware.
This area covers topics such as IT services, IT architectures, IT management and telecommunications.
Sustainable insulating material for shipping temperature-sensitive products
Researchers at the Institute of Natural Products Engineering at TU Dresden have developed an insulating material made from recycled paper for shipping temperature-sensitive foods and medicines. As part of a…
Helium nuclei at the surface of heavy nuclei discovered
Research team confirms a new nuclear property predicted by theory Scientists are able to selectively knockout nucleons and preformed nuclear clusters from atomic nuclei using high-energy proton beams. In an…