READERS PLEASE NOTE: This article was published
A Ryerson professor has developed software that will allow computers to read your thousands of tweets and understand how much you hate your phone company, or how upset you might be with Ryerson for the various times it didn’t call a snow day when it should have.
Denote is a software platform that gives computers the power to recognize negative or positive feelings being expressed on social media by actually understanding the semantics — the meanings of the words.
Ebrahim Bagheri, an assistant professor in the department of electrical and computer engineering at Ryerson, and a team of graduate students developed Denote at the Laboratory for Systems, Software and Semantics (LS3). It could be a valuable tool for companies that need to know what the public thinks of them but don’t want to sift through millions of tweets or look for specific keywords.
“If I was a company like Verizon or Rogers,” Bagheri said, “and I want to know what customers think about me, it wouldn’t really help me to look through Twitter or Facebook and just count the number of people saying good things or bad things.
“What I want to know is, ‘what is the general feeling about my services?’ In order to be able to understand this, I need the semantics behind those tweets or posts.”
Denote can also enhance the ability of searches by identifying the intentions behind what is searched based on the semantics. So instead of just searching for pages, tweets or posts with the keywords typed, Denote figures out what the intention is behind what you’re searching and looks for anything that might relate to that.
“If he had enough time, a human being could read all the documents, these tweets and Facebook posts,” Bagheri said. “But we don’t have that time, we don’t have that person. So now we say, let’s take it to the next level, let’s help the computer understand these documents and once we ask them semantic questions, they would be able to answer them.”
Bagheri and his students are now in the final stages of optimizing the software, which will take up to two months. But once it’s finished, Bageheri said, any application that requires computers to actually understand words could be built based on Denote’s technology.
This story also appeared in The Ryersonian, a weekly newspaper produced by the Ryerson School of Journalism, on Feb 4, 2015.