'iTunes approach' could aid scientific discovery

 作者:沃白     |      日期:2019-03-02 02:03:00
By Tom Simonite Software that creates links between scientific papers, based on a smart analysis of their content, is being developed in the UK. It mimics the way online retailers like Amazon and iTunes make product recommendations. “Those sites have services that say ‘if you bought this, you might be interested in that,'” says Henry Rzepa, a chemist at Imperial College London, UK, who developed the software. “We were inspired by that model.” Retail sites use other shoppers’ buying and browsing habits to generate recommendations. Rzepa’s software uses a slightly different method to create lists of related papers to accompany those cited by the author. He worked with Omer Casher from drug company GlaxoSmithKline’s Clinical Imaging Centre, also in the UK, to create the web application – called SemanticEye. The site is designed to analyse technical chemistry papers. The first version searches a paper for known chemical compounds and uses a simple understanding of the relationship between these compounds to recommend papers that have been added to its database. The same principle could be used to analyse other key elements of a paper, and the researchers are working on a version that incorporates chemical reactions as well. They say it could also be applied to other scientific disciplines. “It can be used to suggest articles similar to the article you are looking at, a bit like iTunes does with music,” explains Rzepa. He is also testing a way to automatically add information from journals to web pages. For example, a page concerning a particular compound would be automatically updated with the latest relevant papers added to SemanticEye. In the future, the system could add links to conference presentations, blogs and other materials. SemanticEye is an example of the ‘semantic web’: an effort to enable machines to process electronic documents according to their meaning. “Because computers don’t get bored and can work so much faster, they can rescue humans from information overload,” explains Rzepa. “The potential is enormous.” He believes the semantic web will speed up scientific progress. “Almost all the great leaps in science have an element of serendipity,