Earth Science Gets Into the Semantic Web Act

By SemanticWeb.com on January 15, 2009 3:21 PM

Jennifer Zaino
SemanticWeb.com Contributor

When it comes to the semantic web, the life sciences community has been a leader in leveraging its capabilities for research purposes. But now specialists in the field of earth science also have a chance to reap the benefits of semantic web technologies. Research scientists at the University of Alabama in Huntsville have developed Noesis, a new semantic web search engine aimed at helping scientists who study the environment be more precise and efficient at searching for the scientific data they need.

Noesis is a Greek word meaning cognizance, or understanding. “The idea is we are trying to see if we can find a better way to search from earth science data and services that process those different data sets,” says research scientist Dr. Rahul Ramachandran.

The search engine uses its ontology in multiple ways. “It helps the end users define their query using the notion of query expansion, where suggestions are given from the ontology,” he says. So, anyone typing in the word “cyclone” in the web-based tool would get the benefit of Noesis assuming that the person is there to research among atmospheric resources, and it will provide relationships for the different concepts related to cyclone, such as synonyms like hurricanes or the physical parameters that tend to be associated with these events. Users can then refine their queries by selecting these additional terms.

“The other part is it goes and looks at multiple resources that have already been collected as the most important ones based on our feedback from scientific domains,” Ramachandran notes.

Ramachandran and principal designer Sunil Movva and their team involved scientists from among multiple earth science disciplines when building the project, to get their input on the sites they normally search. The normal process for many earth science researchers is to Google for a term, but it’s a flawed approach that often leads to irrelevant results, Ramachandran says. Google doesn’t really understand the context of the search, and with so much data out there, it’s very difficult for scientists to narrow those searches down to the right archives.

“With the ontology you can basically force the context,” he says. Also, the data scientists really need may be completely missed by these searches, since with some scientific publications and catalogues users must go directly through their data portals to access information. “We query the data directly through an API,” says Movva, avoiding that step.

Semantic mediation across different data catalogs using the ontology is part of the secret sauce. Ramachandran notes that a lot of effort goes into meeting with earth scientists to create agreement on different concepts and their relationships to inform an ontology. The researchers leveraged the Jet Propulsion Laboratory’s Semantic Web for Earth and Environmental Terminology (SWEET) framework as a building block for the ontology. The initial version of the tool covers atmospheric science, and that’s been followed up by one that substituted an ontology for coastal ecology.

“The idea is that eventually, when different domains develop their ontologies, you should be able to combine them all together and then be able to do this broad search,” says Ramachandran.

The system can be adapted to any branch of science just by switching out the ontology, and anyone can freely use the web tool. One organization that currently is leveraging it is The Federation of Earth Science Information Partners (ESIP). The consortium of more than 90 organizations that collect, interpret, and develop applications for remotely sensed Earth observation information, has adapted the tool for its own search system, says Ramachandran.

Comments