4 Emotion Detection API’s You Need to Try Out
Social Media Week is a leading news platform and worldwide conference that curates and shares the best ideas and insights into social media and technology's impact on business, society, and culture.
Social media and online review sites encourage a ton of writing about your business or brand. With so much valuable data hidden in plain sight, you can’t let it just sit there! Unfortunately, with more and more text generated every moment, the task of sifting through it all can be incredibly daunting. Thankfully, artificial intelligence and machine learning can help you mine large samples of unstructured data easily with emotion detection and sentiment analysis.
What is emotion and sentiment analysis? It’s an artificial intelligence’s ability to scan a face or text and determine. We often think of robots as calculating entities incapable of reading emotion, but that’s an outdated concept—thanks to machine learning, AI is picking up on the nuances of language.
This means with their help, you can quickly find the good and the bad that people are saying about your brand, in scarily specific accuracy. Whether you want to build a happy online community, improve your customer service efforts or monitor discussion surrounding a brand, there are lots of emotion detection tools that can help you out.
How Emotion Detection Works
There are two main strategies for emotion detection: facial recognition and semantic analysis. Facial detection analyzes facial expressions in video and photos, detecting microexpressions which determine common emotions such as surprise, joy, anger, sadness, disgust and more. These powerful algorithms can detect expressions by plotting points on a face and reading their relationships to one another, with the help of facial databases.
The second type of machine learning-based sentiment analysis—and one you’ve likely encountered online—is semantic analysis. Semantic analysis includes algorithms that detect emotion in language, whether it be speech or in writing. This is achieved by ranking whether keywords in a text are positive or negative in connotation, through which an overall tone emerges. Algorithms may detect multiple examples of tone in a single statement, offering a comprehensive look into what the speaker or writer is thinking.
Use Cases For Emotion Detection
There are many great use cases for sentiment analysis in social media, which can be a great boon for marketers and community managers. Examples of use cases include:
- Gauging sentiment tied to a brand in real time across social networks (semantic)
- Detection of bullying or abuse in a community (semantic)
- Campaign evaluation (semantic, facial)
- Large-scale reaction testing to products (semantic, facial)
- Customer service—for example, prioritizing angrier customers for speedier service (semantic)
- SEO—for example, analyzing content of page (semantic)
The facial recognition market is wasestimated a $2.77 billion industry in 2015, and is expected to rise to $6.19 billion in 2020. And with the rise of artificial intelligence and chatbots in the customer service and e-commerce, you can bet semantic analysis will grow of similar importance.
How Can You Make Use of Emotion Detection?
There are a lot of SDK’s and API’s you can grab to make use of machine learning-based sentiment detection in your apps, services and business. With their help, you can step up your marketing and community monitoring game by taking advantage of the latest trend in big data analysis.
Let’s have a look at a few stellar examples.
1. Speech and Text Sentiment Detection with Project Oxford
But Project Oxford isn’t just about facial detection; there are plenty of tools for semantic analysis as well. Its Text Analytics API scores a text on a positive-negative spectrum, giving you insight on user sentiment in whole blocks of text. In addition to scoring sentiment, it highlights key phrases and words behind the sentiment.
For example, if you find many users complaining about your product or service, you can check at a glance specifically what it is they’re unhappy with—and can scan to see if there are similarities between those experiences. Play around with it to see for yourself here.
2. Using IBM Watson’s Emotion Detection for Messaging
Another high-profile API for machine learning emotion recognition is IBM Watson’s Tone Analyzer. That’s right: Watson is good for more than just brainy tasks like answering trivia and playing chess. The tech is excellent for assessing tone in customer service scenarios, as well as for corporate messaging. Let’s imagine a situation in which you might want to use Watson:
Let’s say your brand is in the middle of a big scandal, and you need to put out a statement fast. First, you can use the Tone Analyzer to see how critics are feeling: are they sad? Disgusted with your actions? Angry? Once you’ve got an idea, you can construct your statement to respond. But before you hit “publish,” have Watson analyze your statement to make sure you’ll make the right impression.
Watson can read various social tendencies like agreeableness, openness, conscientiousness and more. These are three tendencies you may want to come across in your message, so check with the Tone Analyzer to see how your statement scores. The tool will also tell you whether the statement is confident or tentative. Try out the tool here.
3. How Bitext’s Sentiment Detection Achieves Customer Satisfaction
Bitext’s API can read eight languages, with more on the way. Like Project Oxford, Bitext measures sentiment and categorizes the content of the text you analyze, with data that’s easy to export to Outlook or whichever data management tool you prefer.
In their demo video, Bitext shows how a restaurant can benefit from these categorized analytics. By sourcing online reviews for data, they find that the “food and drink” category measures highly in negative sentiment. Looking deeper into the category, they find that cost is the main factor that customers are unhappy with.
You can imagine how useful the tool can be for identifying common issues in customer service, helping to streamline your efforts in dealing with criticism and customer satisfaction. And since customer satisfaction is the battlefield where brands will compete this year, it’s worth getting a leg up with big data.
4. Smart Moderation’s Machine Learning-Based Community Moderation
We’ve discussed how sentiment detection and machine learning can help in various aspects of customer service, but we’re forgetting a crucial aspect of social media marketing: community management! Enter Smart Moderation, a machine learning-based tool that’s capable of automatically moderating communities across all your typical social networks with a high accuracy rate. With the help of its API, you can integrate the software into any digital platform. It checks for spam, abusive language, profanity and more—because it’s adaptive, you can tailor it to your individual needs.
So, why include it in our roundup of emotion detection tools? The software’s smartest asset is arguably its ability to read sentiment: for example, you may want to moderate against profane language, but keep phrases like “[your product] is f***ing awesome!” With more primitive AI, that phrase would be banned simply because of the presence of the f-word; but with an intelligent, machine learning base, the tool is able to recognize it as a compliment. You can start a free trial here.
Learn the latest trends, insights and best practices from the brightest minds in media and technology. Sign up for SMW Insider to watch full-length sessions from official Social Media Week conferences live and on-demand.
Want to write for Social Media Week?
We're looking for individuals around the globe to contribute articles on marketing, media, technology, and more.