May 12, 2006 | Sharing their passion for the Semantic Web at the Bio-IT World Life Sciences Conference + Expo were several prominent leaders of the movement.
Providing a clinical perspective was Tonya Hongsermeier of Partners Healthcare, who said the Semantic Web is necessary for knowledge management because “the market is balking at healthcare inflation.” She said eight datapoints are now kept for each patient (age, height, weight, etc.), but with new tests and personalized medicine, the number could balloon to 3,000. The Semantic Web can reduce the cost and liability of knowledge management, said Hongsermeier, and is a critical component of translational medicine going forward.
Eric Miller, Semantic Web lead at the W3C, spoke of the huge growth of interest in recombinant data, which integrates and reuses data across artificial boundaries. The use of uniform resource identifiers (URIs) in the Semantic Web makes data more machine-readable, freeing it from the application in which it was created to be used in other applications. The use of resource description framework (RDFs) lets users link objects in relational databases that they could not link before.
Explaining the difference between current Web applications and the Semantic Web, Teranode’s Matt Shanahan said the Semantic Web, in effect, adds “lenses” that let you visualize the data in different ways. In addition, Semantic Web applications incorporate multidimensional links, pop-up boxes with related data, and annotations, in contrast to the regular Web’s plain hyperlinks.
There is a great need for data integration in biology, said Teranode’s Eric Neumann, and the Semantic Web provides the flexibility to “plan for change.” He stressed the importance of the power of links — the “network effect” that the Semantic Web provides, as well as the lower reliance on complex middleware. Within a Semantic Web framework, data are not a string, but a URI, and using the same URIs lets you “stitch together” different data from different sources.
Oracle’s Susie Stephens and Northeastern University’s Ken Baclawski offered a programming perspective. Stephens touched on the importance of RDFs, which she said are complementary to Oracle’s databases. Baclawski addressed the question: How can you use the logic-based Semantic Web when a lot of the information you’re searching for is uncertain? The answer, he said, is in Bayesian networks, which are graphical techniques, and graphics is also an important feature of the Semantic Web.
Asked if the Semantic Web was not reminiscent of the hierarchical databases of the 1970s, Stephens agreed that they are similar, but the big difference is today’s computer infrastructure is much more powerful, allowing the Semantic Web to scale.