Research

No
Julian Brooke

Linguist aims to tap cyber sentiment

July 23, 2009

Document Tools

Print This Article

E-mail This Page

Font Size
S      M      L      XL

Related Links

The web is brimming with commentary and criticism and Julian Brooke (above) thinks it would be useful if we could collectively survey this information to determine its underlying sentiments.

The linguistics graduate, who earned his master’s degree in June, envisions a web crawler capable of creating a "sentiment" map that could be tuned for various products or issues.

"It could be a publicly available map and you could watch, day by day, to see what people think about things," he says.

The trick is to figure out how to extract sentiment from text, a topic he explored in his master’s thesis. To date, researchers have been using a basic knowledge of how language works to determine which words and phrases are positive or negative.

Brooke, who earned a BA in symbolic systems—computing, psychology and linguistics—from Stanford University, has taken the research a step further.

He has been examining linguistic structures more complex than just words, such as clauses, paragraphs and adjective/noun pairs to determine a text’s underlying sentiment.

It’s a computational problem that also interests Microsoft, which is why Brooke has a three-month summer internship as a researcher with the software giant’s Internet portal, MSN.

"It will allow me to find things out about the scope of the problem and to understand the problem better," he says.

It’s all quite a switch for a fellow who spent six years in China, first teaching English and then learning Mandarin and becoming a translator.

After Microsoft, Brooke is off to the University of Toronto to earn a PhD in computing science. There, he plans to delve further into the computational wizardry behind polling people’s feelings in cyberspace.

Comments

Commenting is closed
Comment Guidelines
Search SFU News Online