Efficiency. Young modern businessman analyzing data using computer while sitting in the office

Ironically, one of the most common phrases in any language must be, “you don’t understand what I’m saying.” While most of us, to a more or less degree, communicate most of the time to most people, breakdowns are evident everywhere. From dueling political parties, to parents and teenagers, language can seem more like a divisive factor rather than a way to share thoughts.

When you think about language comprehension, it’s a wonder that we understand each other at all. Communication 101 tells us that the process involves a sender-message-receiver, and if we reduce transmission noise all should be well. However, things are much more complicated than that when the sender and the receiver are human brains and the message is encoded in words. Language comprehension is an extremely complex process with many layers of interpretation.

To begin, understanding any linguistic element requires semantic analysis of the literal meaning of the “text,” the word, sentence or narrative as it is spoken or written. This is more than dictionary meaning, and includes syntax and context in a broader linguistic framework, such as a word in a sentence, or a sentence in a conversation. It relies on determining all the possible references, narrowed by the sender and receiver’s knowledge of the world, context in the physical world (where it is said), and social context (who says what to whom).  So, when your teenage daughter says “I’m dead, it’s tragic he’s salty,” you know she’s really very much alive from the physical context and decipher that she has a certain attitude about another person from the linguistic context (with help from the Urban Dictionary).

Often, words in isolation are not enough to determine meaning. Further interpretation involves understanding the intention and sentiment of the speaker, which takes all of the previous analyses and attributes a positive or negative attitude to the words expressed. Seems simple, but humans can be sarcastic and say just the opposite of what they mean. Other times we say what we think others want to hear or what makes us look good. For example, your daughter states that a situation is negative, but her meaning is that she feels positive or indifferent. But we’re not done yet.

Specific emotions can be determined by a more detailed affective analysis, but this is not always easy. First, it takes identifying the possible range of emotions expressed in language, and eminent psychologists and sociologists have argued about this for decades.  Second, it takes knowledge of individuals in a social context at any particular point in time. Even if you know your daughter’s emotional landscape well, you may have difficulty discerning that she’s actually upset about her friend.

Assuming that interpretation is successful up to this point, there’s still the additional problem of determining how someone is motivated to act according to the emotions they’ve expressed. If this seems difficult on an interpersonal level (what does your daughter want to do about her friend?) consider how impossible it could be for predicting the behavior of larger populations, like consumers, voters or a public confronted with a pandemic.

Fortunately, Cognovi Labs has developed a way to help marketing agencies, political pollsters and public health officials through this process.

Cognovi’s innovative approach starts with powerful semantic processing AI research, generated over eight years at Wright State University’s Kno.e.sis Center and funded by the US Air Force, the Department of Energy, and the National Science Foundation. Their proprietary methods mine textual data in social media texts, discussion forums, chat rooms, transcribed conversations or intra-company communications. If these sources don’t provide the appropriate data, their Dynamic Diagnostic Interview techniques will. This produces the salient words and phrases used by a certain population and their semantic analysis.

AI strategies enabled by machine learning are then applied to these data, which extract expressed emotions such as joy, anger, disgust, fear, sadness and surprise. According to Beni Gradwohl, Cognovi’s co-founder and CEO, they go further than sentiment analysis by focusing not on what people say or how they say it, but on what they are feeling when they do. Importantly, their algorithms identify the context of emotional expressions, all measured in real time and at scale.

While the data analysts and AI scientists on Cognovi’s team produce impressive results, their clinical and cognitive psychologists take them to the next level by predicting motivation and decision-making that influences behavior. In addition to prediction, they can also prescribe efforts that can be taken to influence decision making and ultimately behavior.

Cognovi’s successes offer concrete validation of their methods. For public health efforts, they devised a Covid Panic index and Vaccine Confidence Tracker. Recently, one of their studies found that promoting certain eligibility guidelines was, in fact, detrimental. Participating in the Veterans Administration Tech Sprint, they developed a roadmap for identifying and assisting at-risk veterans with PTSD. For healthcare marketing, they prescribe the best time to launch a life-saving medication to ensure acceptance by doctors and patients. Financial advisors and investors benefit from Cognovi’s market insights, and political campaigns from their election predictions.

There’s also the vital research they’ve conducted on football fans’ attitudes toward a NFL quarterback pick and fashion trends during the pandemic (no, we weren’t all sitting around in pajamas).

So now we’re talking. We can only hope that Cognovi can determine what teenagers really feel about their besties … and advise parents on helping them navigate adolescent storms.