For researchers, gathering information is becoming easier. Now, academics have a more comprehensive means for collecting data.
Cornell University Computer Science Professor Claire Cardie delivered a speech on noun-phrase coreference resolution yesterday afternoon in the Wu and Chen Auditorium of Levine Hall.
Noun-phrase coreference resolution is a process by which nouns or phrases that refer to the same entity -- such as "New Jersey" and "the Garden State" -- are recognized and compiled from hundreds or even thousands of different documents.
Noun-phrase coreference resolution takes this process of noun-phrase matching and executes it on a much broader scale.
For example, identification of "Garden State" over a thousand documents would yield an extensive table of synonyms referring to New Jersey.
Fernando Pereira, chairman of the Computer and Information Science Department at Penn, elaborated on the process.
"It's the same way you might write an essay and collect a bunch of facts," he said. "Only you're collecting that information much more quickly."
In addition to defining the term, Cardie explained the different methods of executing noun-phrase coreference resolution in her speech.
Cardie emphasized these methods as "a couple of classes of techniques that are very different," some of which work better under certain conditions.
The Cornell professor attributes her passion for language to "Introduction to Natural Language," a class she once took in college.
"That's what changed my mind," she said. "It revealed the notion that you can combine computing with trying to understand language."
At Penn, noun-phrase coreference resolution has greatly aided CIS Department members in the area of medical research.
"There are many articles on medical science, and researchers cannot keep up with them," Pereira said. "Noun-phrase coreference resolution allows us to mine huge texts for specific medical information."
At the end of her lecture, Cardie opened up the floor for questions.
At the reception following the presentation, the professor mingled with audience members as they engaged in discussion about the progress of noun-phrase coreference resolution.
CIS graduate student John Blitzer focused on the use of machines versus humans to gather information.
"We don't learn in the same way computers are learning," he said. "As a result, with humans and machines the different algorithms function in different ways."
Cardie's speech was part of the Computer and Information Science Distinguished Lecture Series for fall 2003.
