In the age of the internet, there is so much information flying around it can be hard to sort through it all. Now a system that organises information by mimicking the way birds of the same species flock together could help.

Web feed tools such as Really Simple Syndication (RSS) automatically alert a user when new information is added to a favourite website. This provides automatic updates when fresh news stories are posted to a site or when new scientific papers are added to a journal, for example.

But making sense of all these automatic updates can be a lengthy process. So Xiaohui Cui and colleagues at Oak Ridge National Laboratory in Tennessee, US, developed a feed-organising system based on flocks of virtual birds. The system uses a modified version of the software that creates animations of birds in movies and video games.

Smart search

Each "bird" carries a document, which is automatically assigned a string of numbers depending on the words it contains. Documents with a lot of similar words have number strings of the same length and a virtual bird will naturally fly with others carrying documents with number strings of the same length.

When a new article appears, software scans it for words similar to those in existing articles and then files the document into an existing flock, or creates a new one. The team has used the system to categorise online news stories from CNN and the BBC. The next step will be to allow people to click on a bird to display its document.

Species recognition

"In nature, birds use properties like colour or voice to recognise species, but the virtual birds use documents instead," Cui explains. He adds that the system provides an easy way to organise information. "It is also more natural because I don't have to specify how many categories I think there are," he says.

Susan Stepney, a computer scientist at York University, UK, who also studies virtual flocks, says it could beat keyword news searches by coming up with associations that users would not normally think of. But she adds that flocking software can use a lot of computing power, so it might be difficult to sort through large numbers of documents this way.