Thursday, March 19, 2009

storytelling

Physorg | New search engine can be used for creative discovery. When you ask a supercomputer to tell a story, you might not expect a creative outcome – or any. But a group of Virginia Tech researchers are using System X, the university’s supercomputer, to test a new search program that can tell the stories of life – the connections between gene sets, for instance, or the connections between discoveries reported in biomedical articles on the U.S. National Library of Medicine PubMed database.

We are all familiar with search engines that provide a list of hits on the terms we enter. Researchers in computer science and biochemistry at Virginia Tech have created a search capability that they call Storytelling that will discover connections between information that appears dissimilar. It discovers a sequence of events or relationships to create a chain of concepts between specified start and end points. Imagine, for instance, asking for a connection from the concept “traveling in London,” to the concept “places popes are interred.” The Storyteller might postulate “the history of codes” as an intermediary and find the Da Vinci Code – if it existed.

“The stories are pieced together by analyzing large volumes of text or other data” said Naren Ramakrishnan, associate professor of computer science at Virginia Tech who works with life scientists to create software for data mining and information analysis tasks arising in biology. The aim is to help scientists make connections in the complex, burgeoning world of scientific discovery. “Everyday, there are new research results reported in the literature and there are discoveries waiting to be made by exploring connections,” said Ramakrishnan.

“Our minds cannot correlate all available datasets efficiently and with any high degree of confidence without the aid of computational biology,” said Richard Helm, associate professor of biochemistry. “Attempting to find significant correlations within the ocean of online datasets is daunting. However, there may be experiments that have been published in the literature that look at particular subsets of a biological process. The storytelling algorithm links ‘distant’ objects by finding these closer connections and drawing them together in a storyline. Evaluation of these stories can provide hypotheses that can be tested at the bench, potentially resulting in new insights into the role of a particular molecular event in the process you are interested in.”

The design of the storytelling algorithm is modeled after large scale search engines such as Google. Each “node” in System X, an 1100 Apple Xserve G5 cluster supercomputer, is responsible for indexing a portion of the biological literature and the nodes exchange information among each other to help define links and make connections. “Some of our larger storytelling runs process hundreds of thousands of papers and work with up to 200 nodes simultaneously,” said Ramakrishnan.