Searches That Try Scientists' Souls

3 minute read
Greg Crites avatar


Click over to your preferred search engine and type “global warming,” hit enter and wait for the faint tang of smoke curling up from your processor. What’s that? Six billion hits! Let’s get sophisticated and try “global warming” and “Al Gore”. Oh, wait. Google just sent you an email threatening legal action if  you keep instigating "insanely vague" searches.

If you’re a scientist, attempting environmental studies with a focus on atmospheric phenomena, and you type "red tide," it doesn’t really help to see links to redheads.com. In fact, it’s a distraction you don’t need.

This is when you may need Noesis, a new semantic web search engine developed at the University of Alabama in Huntsville and a promising addition to scientific research.

What is Noesis?

Originally developed for scientists working in the field of Atmospheric Science, Noesis' framework could be replicated in other scientific fields of study as a better way to search through the "hidden web" of scientific catalogs that traditional search engines cannot reach. Noesis is using the semantic model: -subject-predicate-object- within the framework of a pre-annotated ontology.

Ontology is a shared vocabulary used to model a domain. More specifically, to model types of objects and/or concepts that exist, as well as their properties and relations. In layman’s terms, Noesis will guarantee scientists and researchers only obtain the most relevant returns on queries. Goodbye redheads.com, hello "why are the fish dying"?

Because the "hidden web" of scientific catalogs is built using a specific vocabulary, the most efficient searches on the catalogs involve using specific terminology from that vocabulary.

Learning Opportunities

To create Noesis, researchers painstakingly annotated specific vocabulary terms with ontologies, creating computer-friendly definitions of the terms that assist the query engine in understanding the concept of the term and its relationship to other terms. This was achieved by using the ontology web language (OWL).  But that’s not all that is involved in tweaking the perfect scientific search.

Noesis employs something they call the Ontology Interface Service (OIS), a SOAP-based web service interface to an inference engine. When a user performs a search, the OIS is also immediately searched for associated concepts. The specific hits and general hits discovered are returned in a tree structure which the user can navigate and refine further. The search also returns synonyms and related terminology. Then, using checkboxes, they can be appended or eliminated from the original query to refine it further.

Whatever Helps the Environment, Helps You

There is almost 1 billion U.S. government web pages. That’s just the U.S. government! The total number of web pages may be larger than the national debt, and that figure is nearly impossible to wrap your head around. Suffice it to say, Einstein would not have been happy choosing “I'm feeling lucky” when doing research. The very disparate and chaotic nature of the web makes it incomprehensible how any type of worthwhile semantic search engine can be implemented web-wide.

But clever people are trying. Noesis, by refining the tools available to researchers and scientists is a welcome compromise available now.