Scientists Developing Web Site Credibility Tools

Researchers around the world are working to create new tools that can quickly ascertain a particular Web site’s credibility.

The need for such a tool has increased in recent years with the surge in the number of Web sites offering health advice, political viewpoints and other user-generated content, said researcher Andreas Juffinger at the Know-Center, an Austrian technology research center.

Juffinger and his team are developing a program that analyzes and ranks blogs as “highly credible”, “average credible” or “little credible,” he told the AFP.

The tool works by comparing statistical metrics, such as the distribution of particular words, of a blog post with those of news articles on the same issue from mainstream sources previously deemed credible.

“It has shown promising results, we think we are on the right path,” Juffinger said during a global World Wide Web conference in Spain.

“It has to be automatic because it is not possible for customers to label and read all these blogs.”

In Japan, researchers are working to develop a program that continually mines the Internet for different viewpoints on a particular topic, and then presents them to users along with a “statement map” clarifying how the different opinions are related.

“We really believe that ‘statement maps’ can help users come to conclusions about the reliability of a Web site,” Koji Murakami of the Nara Institute of Science and Technology, one of the researchers working on the project, told the AFP.

The number of Web sites worldwide has jumped from just 500 in 1994 to tens of millions today, according to data from Microsoft.

Wikipedia, one of the Web’s most-visited sites, is an online encyclopedia with open-source software such that anyone reading a subject entry can edit it.  This feature has created doubts about the site’s credibility ““ suspicions that surfaced again in January when an entry on Senator Edward Kennedy was changed to erroneously report his death following a seizure he suffered at President Obama’s inauguration luncheon.

However, despite these occasional incidents, Wikipedia has grown to have some 2.6 million articles in English alone since its founding in 2001.

Italian researchers are developing an algorithm to verify the integrity of Wikipedia entries by assigning quality scores to each article and contributor.

“Preliminary results demonstrate that the proposed algorithm seems to appropriately identify high and low quality articles,” wrote the researchers, led by Alberto Cusinato of Italy’s University of Udine, in a paper presented at the Web conference.

In 2005, the British journal Nature reviewed a wide range of scientific entries on the Wikipedia site and in the Encyclopedia Britannica.  The team concluded that although Wikipedia entries were often poorly structured, there were few differences in accuracy between the two sources.

On the Net: