Here at Echosec, we help organizations find and filter billions of social media posts every day. When we geofence a location on the map, we see all of the posts coming from that location which helps us grasp the entire post context. We see the location, the time & date of the post, whether or not there are images associated with much of the text, and because we’re human beings, we are much better than computers at detecting nuances like sarcasm. With these advantages, the overwhelming sentiment of people in that location is often very obvious.
Echosec helps organizations understand the sentiment towards their organization or brand, but we do so in a different way than the current definition of sentiment analysis. We think our way is better, here's why.
What is Sentiment Analysis?
Currently, sentiment analysis is defined as the interpretation of text in social posts, to understand the feeling or attitude of the people who wrote them.
Sentiment analysis, (also known as opinion mining or sentiment tracking), is a feature of many social platforms on the market today. The feature usually consists of a simple thumbs up/thumbs down, or smile/frown to indicate positive or negative sentiment. It’s been around for many years now, and the technology behind it hasn’t really changed much!
What’s good about it?
These days, the whole world is wearing their hearts on their news feeds. It's no secret that the past decade has shown a global increase in public displays of opinion, and this shift has created a new way for brands to learn about their current customers, acquire new ones, mitigate PR disasters and stay reputable. It’s a way for companies to turn our oversharing and humble-bragging into business intelligence, improving their reputation and their bottom line.
Where Does Sentiment Analysis Fall Short?
Anybody who’s ever broken up with someone via text message understands the struggle. Text is often misunderstood, diluted, and misleading, especially when you’re working with 280 characters or less.
At best, text analysis as it stands today, offers a neat way to see the most rudimentary interpretation of the emotion contained within a post.
At worst, the analysis tool fails to perceive tone - such as sarcasm, and interprets the emotion in the complete opposite way from which it was intended.
"I *love* mornings."
with a picture of a sunrise means something VERY different to,
"I *love* mornings"
with a picture of coffee grounds all over the kitchen floor.
Sentiment analysis has a long way to go in its evolution.
How could it be better?
There's more to the story than words alone...
Understanding sentiment will increasingly deliver incredible value for your brand. We think the next iteration of sentiment analysis will factor-in written emotions, such as surprise, disgust, anger, confusion, as well as meaningful context such as location, time, and current events. This will give organizations an incredible depth of knowledge about their customers.
With advancements in machine learning, and the Internet of Things, we are starting to get a much clearer and more accurate picture of the conversation. More context and a broader set of emotions give us the power to turn sentiment analysis from a simple +/- gauge to a whole new space to explore.
To sum it up,
Sentiment analysis as it stands today only delivers a small piece of the emotional puzzle that is true human sentiment.
To truly understand the sentiment around your brand, consider text, imagery, time, place, and overall context of social media posts and other online data.