Wednesday, December 27, 2006

Tim Berners Lee on Semantic Web

The Economist has an article of Time Berners Lee titled "Welcome to the Semantic Web" (subscription req) on their "The World in 2007" section

"Data integration will be the web’s next leap forward. The most exciting discoveries will come from the serendipitous combination and integration of data drawn from diverse sources."

"The Semantic Web will derive its power in a similar way, but through the linking of data rather than documents. To appreciate the need for better data integration, compare the enormous volume of experimental data produced in commercial and academic pharmaceutical laboratories around the world with the frustratingly slow pace of drug discovery. Life-science researchers are com­ing to the conclusion that in many cases no single lab, library or genomic data repository contains the informa­tion necessary to discover new drugs. Rather, the infor­mation needed to understand the complex interactions between diseases, biological processes and the vast array of chemical agents is spread across disparate databases, spreadsheets and documents. As a result, progress towards better drug discovery de­pends on technologies that enable sharing and integration of data, as well as on changes in institutional practices in order to allow exploration of the links in these data."

This is not to suggest that all pharmaceutical companies sim­ply free their data, but rather that they explore more flex­ible licensing models that allow greater value to be created through the combination of their own intellectual prop­erty and that of others. For this sort of integration to happen, an essential technical step is to publish the data using Semantic Web standards (RDF, OWL, SPARQL), and to link them together with definitions of the terms used to express the data. For example, when publishing experimental results about the behavior of a particular chemical in a larger biological process, one must indicate which vocabularies are being used to describe the biological pathway and the chemi­cal. Then, when someone else wishes to integrate those data-for example, with other experiments documented in the research literature-that person can use that same vocabulary to match article keywords to chemical names. via JLReis

No comments: