If you’ve ever found yourself scratching your head and wondering if your uncle’s outlandish Facebook post was meant to be satire or not, you’re not alone. It can be tough to understand the subtext of written language, particularly when it comes to social media posts that are often meant to be written in a pithy and concise manner.
And it’s not just tough for humans — it’s also tough for computers. Even though there are plenty of language models nowadays that are capable of undergoing processes resembling genuine understanding of human language, it remains extremely difficult (and perhaps even impossible) for them to understand the deeper meaning associated with a particular text.
That could be changing soon though. A team of natural language processing (NLP) specialists at Purdue University are making efforts to use machine learning techniques to train artificial intelligence models to understand and identify the subtext of a particular text. “The motivation of our work is to get a better understanding of public discourse, how different issues are discussed, the arguments made and the perspectives underlying these arguments,” said Dan Goldwasser, an associate professor of computer science at Purdue who worked on the study.
“We would like to represent the points of view expressed by the thousands, or even more, of people describing their experiences online,” he added. “Understanding the language used to discuss issues can help shed light on the different considerations behind decision-making processes, including both individual health and well-being choices and broader policy decisions.”
As much of online communication relies on the fact that readers are aware of the context in which it occurs — after all, many of the memes one comes across on Twitter might seem completely baffling if you were to come across them on a platform outside of the digital sphere — Goldwasser and his lab are currently working to use NLP methods to gain a better understanding of the ways in which machines can learn how to factor context into their analysis. As such, the team focuses on the ways in which language is used on social media, in traditional news media as well as in legislation, so that each form of language can be associated with a specific context.
“In many of the scenarios we study, progress relies on finding new ways to conceptualize language understanding, by grounding it in a real-world context,” Goldwasser said. “Operationalizing it requires developing new technical solutions.”
