Seeing Artificial Intelligence in Real Life as it Learns to Detect Sarcasm Online

You that artificial intelligence has arrived when it learns to recognize sarcasm on social media. After all, this is something that everyone must know to survive in today’s world.

Of course, from a technological standpoint, this is a fantastic accomplishment. Researchers from the University of Central Florida really have created a sarcasm detector using artificial intelligence.

Jumping into the social media community

Society today has wholly immersed itself in the world of social media. Practically every thread of modern communities exists there in some form or fashion.

While individuals conduct various actions among themselves, companies are using it to promote their products and their brand, and influencers use it to promote their causes. Not only that, but marketers pay billions every single year to advertise on this medium. And, of course, they do so because of the massive audience that social media attracts daily.

Understanding the role of artificial intelligence

UCF computer scientists fully understand the value of creating artificial intelligence that can interpret dialog on social media. They know that every marketer and company under the sun will be interested in such an algorithm.

They understand that recognizing sarcasm will go a long way in correctly evaluating feedback – which is critical to many businesses. When you consider that companies have long sought opinions online because people are free to express their honest opinion and the fact that sarcasm is used excessively in our culture, then it’s a win-win artificial intelligence project.

Structuring the artificial intelligence software

Creating this algorithm was quite a challenge. Whenever verbiage is analyzed in a logical sense, sentiment comes into play. The goal was to recognize the emotion associated with a given block of text—and there are only three possible outcomes: positive, negative, or neutral.

The findings of the research team were posted in the journal Entropy.

What the team did first taught their computer model how to recognize patterns that are associated with sarcasm. Next, they taught it how to correctly identify words in specific sequences that would most likely indicate sarcasm. They were able to teach the AI this by feeding extensive data sets to it and then checking the accuracy of its results.

“The presence of sarcasm in text is the main hindrance in the performance of sentiment analysis,” pointed out by Assistant Professor of Engineering Ivan Garibay. “Sarcasm isn’t always easy to identify in conversation, so you can imagine it’s pretty challenging for a computer program to do it and do it well. We developed an interpretable deep learning model using multi-head self-attention and gated recurrent units. The multi-head self-attention module aids in identifying crucial sarcastic cue-words from the input, and the recurrent units learn long-range dependencies between these cue-words to better classify the input text.”

Functions of the research team

The research team also included a computer science doctoral student named Ramya Akula, who started working on this project under a DARPA grant that supports the Computational Simulation of Online Social Behavior program.

“Sarcasm has been a major hurdle to increasing the accuracy of sentiment analysis, especially on social media, since sarcasm relies heavily on vocal tones, facial expressions, and gestures that cannot be represented in text,” urges Brian Kettler, a program manager from DARPA’s Information Innovation Office. “Recognizing sarcasm in textual online communication is no easy task as none of these cues are readily available.”

Garibay’s Complex Adaptive Systems Lab (CASL) has been studying challenges like these for some time. CASL happens to be an interdisciplinary research group that has dedicated itself to understanding complex phenomena like innovation ecosystems, global economy, global information environment, sustainability, and social and cultural dynamics. CASL scientists evaluate problems like these using network science, cognitive science, data science, complexity science, machine learning, deep learning, and even the social sciences.

“In face-to-face conversation, sarcasm can be identified effortlessly using facial expressions, gestures, and tone of the speaker,” Akula points out. “Detecting sarcasm in textual communication is not a trivial task as none of these cues are readily available. Especially with the explosion of internet usage, sarcasm detection in online communications from social networking platforms is much more challenging.”

With each passing day, we feel more and more impact of artificial intelligence on society. Who knows what comes next?