Sentiment Analysis on the Cutting Edge


One of the much-vaunted benefits of text and social network analytics is the ability to determine how people feel. Whether by assigning weights to positive and negative terms, scanning for smiley emoticons, or using facial-recognition algorithms, sentiment analysis is garnering plenty of attention.

In fact, it's going to be the main focus of a two-day conference this week in New York. The Sentiment Analysis Symposium kicks off Wednesday with half-day workshops on technical and business tracks, and it continues Thursday with a packed agenda of speakers and topics.

As we've discussed here, many sentiment analysis efforts combine human intelligence with machine learning to train algorithms and improve their accuracy. Jason Baldridge, an associate professor of linguistics at the University of Texas, will cover this in his "Practical Sentiment Analysis" tutorial Wednesday. In the description, he says today's technology can take efforts beyond traditional "opinion mining" and unlock more insight from semi-supervised learning and nontextual information like social networks, sensor feeds, audio, and video.

Thursday's schedule involves a rapid-fire succession of short talks. I'm hoping to attend the afternoon sessions, at the very least. Here are some topics in which I'm especially interested:

  • MIT Media Lab professor Rosalind Picard will discuss the latest in facial recognition machine learning, which, she says, has become "highly accurate" at differentiating between frustrated and delighted smiles.

  • University of Maryland professor VS Subrahmanian will talk about "Forecasting the Spread of Sentiments and Emotion in Social Media." In the course description, he says anger and fear can be traced via Facebook and Twitter to the benefit of retailers, online services, and politicians. "In this talk, we will use a major forthcoming international election as a colorful case study of how to track sentiment and emotion on any given topic, make forecasts about how far those sentiments and emotions will spread, and suggest ways in which to maximally influence the outcome to be consistent with the needs of client." This talk about influencing outcomes is reminiscent of the social physics theories put forth by MIT data scientist Alex Pentland. It seems awfully sinister, but maybe it's just good old marketing and public relations, with a little more data thrown in.

  • Dell's Shree Dandekar will talk about using enhanced text analytics to generate "social ROI" in real-time. This session is scheduled for just 10 minutes, so I'm hoping it will blow my mind.

  • Yuval Mor, CEO of Beyond Verbal, will talk about analyzing vocal intonation to determine emotion. In his course description, he discusses implications for machine interfaces that understand how humans feel, and he promises to share best-practices for incorporating emotion detection into voice solutions.

These kinds of cutting-edge applications seem almost like the stuff of science fiction. Can I program my phone to screen out angry callers? Can I craft a message that's guaranteed to sway people's emotions? How far out are these innovations? I'm hoping to get the answers this week.

Members, are any of you going to the conference? What do you think about sentiment analysis?

— Michael Steinhart, Circle me on Google+ Follow me on TwitterVisit my LinkedIn pageFriend me on Facebook, Executive Editor, AllAnalytics.com

Related posts:

Michael Steinhart, Contributing Editor

Michael Steinhart has been covering IT and business computing for 15 years, tracking the rising popularity of virtualization, unified fabric, high-performance computing, and cloud infrastructures. He is editor of The Enterprise Cloud Site, which won the Least Imaginative Site Name award in 2012, and he managed TheITPro.com, a community of IT professionals taking their first steps into cloud computing. From 2006 to 2012, Steinhart worked as an executive editor at Ziff Davis Enterprise, writing and managing research reports, whitepapers, case studies, magazine features, e-newsletters, blog posts, online videos, and podcasts. He also moderated and presented in dozens of webinars and virtual tradeshows. He got his start in IT journalism at CMP Media back in 1998, then moved to PC Magazine, managing the popular Solutions section and then covering business technology and consumer software. He holds a Bachelor of Arts degree in communications/journalism from Ramapo College of New Jersey.

Biosensors, Robots Replacing Human Guards

In prisons and other security settings, electronic devices are predicting and detecting anomalous behavior with high accuracy and reducing the need for human personnel.

Questioning the Ethics of Learning Analytics

Schools around the country are weighing the benefits and ethical questions of learning analytics.


Re: Sentiment Symposium 2014
  • 3/6/2014 4:55:00 PM
NO RATINGS

Call centers have plenty of aggressive tones. (Money quote)

Voice gathering, pre-processing, emotional analysis, and business implementations.

Re: Sentiment Symposium 2014
  • 3/6/2014 4:50:51 PM
NO RATINGS

"What if Siri or Samsung S Voice understands our emotions?" Beyond Verbal's Dan Emodi introduces a free demo app called Moodies. Looks like it's only for Apple, though.

Re: Sentiment Symposium 2014
  • 3/6/2014 4:42:32 PM
NO RATINGS

Raw vocal intonation analysis - 

It's not what you say, but how you say it.

Vocal intonations - mood, attitude, personality

Re: Sentiment Symposium 2014
  • 3/6/2014 4:39:31 PM
NO RATINGS

Now we're getting into "Emotions in Speech" which I think is going to be an interesting talk. Beyond Verbal is the presenting company.

Re: Sentiment Symposium 2014
  • 3/6/2014 4:35:41 PM
NO RATINGS

I'll admit that this discussion of deep learning is somewhat over my head -- but Pulman cautions that deep learning is going to provide small improvements, not huge ones. It's important to not buy into the hype generated by a 'naive' press. That's me!

Re: Sentiment Symposium 2014
  • 3/6/2014 4:26:43 PM
NO RATINGS

The advantage of 'deep' NLP is that you can train on billions and billions of words, Pulman says. This is driven in part by GPUs and multicore/cluster/high-performance systems.

Re: Sentiment Symposium 2014
  • 3/6/2014 4:19:24 PM
NO RATINGS

Now it's time for - Bleeding Edge NLP: Extract & Analyze Events, Relationships, Emotion, Intent, Identity, and Lots of Other Good Stuff

Stephen Pulman, University of Oxford

 

Re: Sentiment Symposium 2014
  • 3/6/2014 4:10:57 PM
NO RATINGS

An exec from Mattel raises an interesting conundrum: Parents may hate a particular toy, and tweet about their hatred, but then buy it anyway because their children want it. Catherine Havasi is trying to draw an analogy to pet owners and pet food, but I don't think it flies. 

Re: Sentiment Symposium 2014
  • 3/6/2014 4:05:55 PM
NO RATINGS

Catherine Havasi is talking about deriving social ROI from measuring brand impact and refining products.

Re: Sentiment Symposium 2014
  • 3/6/2014 3:56:21 PM
NO RATINGS

Hoskins says Amazon's goal is to know what you're going to buy before you do, and have it on the truck, on its way to you.

The value of social media is in showing what people are saying/doing 'in their natural habitat' as opposed to a captive or opt-in audience.

What about the people who didn't opt in?

At Amazon, we know about the people who didn't buy and why.

<<   <   Page 2 / 7   >   >>
INFORMATION RESOURCES
ANALYTICS IN ACTION
CARTERTOONS
VIEW ALL +
QUICK POLL
VIEW ALL +