Sentiment Analysis for social media and marketing – Digital Surgeons

In 1975, two New York inventors created the mood ring. The ring contained an element that changed color dependent on your mood. It skyrocketed in popularity and such rings are still sold today. Mood rings reflect your mood as much as twitter bots and strippers actually like your personality: not at all. Is sentiment analysis the mood ring of the digital world?

Sentiment Analysis: For those not MIT Educated

For those without a formal education in computational linguistics, let’s take you on a crash course in Sentiment Analysis. Sentiment Analysis is when a computer analyzes information (emails, tweets, blog posts) and identifies emotion attached to that information. Positive or negative classifications are the most basic categories sentiment analysis can tackle. Sentiment Analysis can also be programmed to find emotions like anger, joy and sorrow in pieces of digital information. The science behind it is subjective at best due to the fact that human emotions (unquantifiable) are being taught to a computer (quantifiable). The widely accepted range of success with sentiment analysis is 70-80% accuracy. This less than perfect accuracy makes a lot of sense considering what is being measured; how often would you say that you misinterpret the tone of an email? Maybe, 20-30% of the time?

A computer can do my job for me? Where do I sign up!?

Sentiment analysis is being used by software companies such SAP and SAS to monitor social media. The idea behind the software is that addressing the moods of consumers can be automated rather than interpreted by an actual human. Instead of paying a community manager to muck through the social media swamps daily, a software can do it for them. Much like cars that drive themselves, these solutions are still a few years from being the standard.

This is the part where the helpful robot kills everyone…

While no human fatalities can be attributed to sentiment analysis, it is still far from being a reliable science. The intricacies of human language are often lost during computer learning activities. On microblogging platforms such as twitter, creativity within those 140 is often rewarded and sometimes misinterpreted. A sarcastic tweet of “That was fun,” could easily be interpreted to mean the literal. Sarcasm doesn’t work well without context and much of sentiment analysis relies on the “bag of words” methodology which ignores context. “Bag of Words” simply assigns mathematical values to words and ignores the structure they are placed. Another perfect example of this methodology falling short would be in the phrases “fuck yeah!” and “fuck, yeah.” One is a positive exclamation; one is a negative confirmation.

A Home in the Marketing Magician’s Bag of Tricks

Much like the mood ring of old, maybe sentiment analysis is only made for fun and not reliable insight? In honor of the Olympics, those talking about the Olympic games had their tweets recorded and their emotions analyzed. Those emotions were then tabulated and the Eye of London Ferris Wheel was lit up to reflect the emotions of those talking about the games. By assigning colors to each emotion, the Eye of London represented less of an accurate data visualization installation and more of a novelty to promote the London Olympic Games. Sentiment analysis may be a not quite ready for prime-time player when it comes to reliable statistics and conclusions, but it functions just fine as a method for bi-directional marketing. Over time it will become more of a mainstay in the marketing analytics field, but for now it is doing just fine in the marketing tactic arena.

Are you using sentiment analysis in your marketing? Let us know which tools you love best.

Popular Sentiment Analysis services/tools:

Tools for measuring sentiment are abundant these days in social media. Here’s a combination of free and mixed tools we have played with or use. Looking for more information on what we know about using sentiment analysis for your social media? Get in touch. After all we can’t tell you all the secrets in one blog post.

Lymbix – http://www.lymbix.com/
Sysomos – http://www.sysomos.com/
Social Mention – http://socialmention.com/
Opinion Crawl – http://www.opinioncrawl.com/
Radian 6 – http://www.radian6.com/
Twitrratr – http://twitrratr.com/
Tweettone – powered by Lymbix – http://www.tweettone.com/
Sentimetrix – http://www.sentimetrix.com/
Lexalytics – http://www.lexalytics.com/
Lithium – http://www.lithium.com/
Viral Heat – https://www.viralheat.com/
Gravity – http://www.gravity.com/
Mood Views – http://moodviews.com/
Datasift – http://datasift.com/
Hootsuite – http://hootsuite.com/features/custom-analytics

View this article – 

Sentiment Analysis for social media and marketing – Digital Surgeons