A Canadian graduate student dissatisfied with science coverage on online sites such as Google News and Yahoo News has created a news aggregator especially for scientists.

Michael Imbeault, an HIV researcher at the Université Laval in Quebec, launched his fully automated site called e! Science News (http://esciencenews.com) last month. It has already attracted 300,000 different users, and averages 5,000 visits a day, he says.

News aggregators display headlines and snippets from other media sources, but don't produce their own content. Of the top five online US news sites, three are aggregators — Google News, AOL News and Yahoo News — and only two — CNN.com and MSNBC.com — generate original content. Yahoo and AOL use human editors and source almost all science stories from wire agencies, such as Reuters. Google News uses computer algorithms to aggregate headlines from thousands of news sources, ranking them by how often and on which sites stories appear. Science and technology coverage on Google News, for example, is notoriously devoid of basic science.

Imbeault's site indexes science news sites, clusters similar articles together on the basis of the frequency of word co-occurrence, and then uses Bayesian statistics to automatically assign articles to topics such as astronomy, health and climate. It then ranks them using factors such as timeliness, and the number of sites reporting the same news, which indicates the story's importance. At present, it is limited to around 40 news sources — including Nature News, The New York Times science section and institutional news sites such as NASA, which offer free content for at least a period — but this will be increased, he says.

Imbeault built the site on top of the Drupal open-source content management software. He says that his aggregator will also be improved by moving to semantics-based techniques that better capture the meaning of a text.