As humans, we include a lot of emotional content when we communicate with one another. V.E.R.N. allows us to not only have a better understanding of the emotions being conveyed but also allows for us to develop tools to help us respond. V.E.R.N. results include a confidence rating on a scale of 0 – 100% of the presence of a given emotion. With this given score, we can provide feedback to a user or system and guide them in their response.
Below are a few examples of how V.E.R.N. could be used within a given setting written in Python for simplicity’s sake. The score being compared is just for sample purposes to demonstrate the logic you could implement within a system. When determining your own thresholds, it will be important to consider your own business workflow.
Medical Patient Pre-Screening
In this scenario, V.E.R.N. is reviewing responses from a pre screening survey for a hospital. These are commonly used to gauge a patient’s mental well being prior to an appointment. For this example we’re going to focus on responding to sadness signals and responding to them.
# Sadness Detection if score > 75: # Strong sadness signals detected # Forward responses to counselor with recommended follow up, notify doctor elif score > 65: # Moderate sadness signals detected # Forward responses to counselor and notify doctor elif score > 50: # Minor sadness signals detected # Forward responses to doctor else: # Minimal sadness signals detected # Basic review recommended
As you can see in the example, V.E.R.N. is not providing a diagnosis but can help triage the responses from patients to qualified individuals. A process which would normally take a human to review prior to making passing on for another review can be directed to the appropriate staff members immediately. In this way V.E.R.N. can be used to expedite communicate and improve patient care.
“I can tell everyone is just waiting for me to die. I’ll never get better.” – Sadness: 80%
In this example, the highest threshold would be triggered. The doctor would be notified of the results and the responses would be forwarded for a counselor to review with the suggestion of setting up a appointment to help the patient.
When an employee leaves a company it can be an emotionally charged time. As a business it’s important to understand the feedback received during and exit interview and be able to respond appropriately. For this example we’re looking for anger signals in the employee’s responses.
# Anger Detection if score > 85: # Strong anger signals detected # Forward responses legal team for potential legal backlash and schedule meeting with employee supervisor elif score > 60: # Moderate anger signals detected # Schedule meeting with employee supervisor else: # Minimal anger signals detected # Basic review of interview recommended
Once again, V.E.R.N. is helping direct communication and responses through appropriate channels. In the case of high anger could indicate the employee was either unhappy with their work environment or they might be upset about leaving the company. In our example, the business reviews the responses with their legal team in order to prepare for any potential legal backlash from the angry employee. Both in that case and when only moderate anger signals are detected, this company requests HR to meet with the employee’s supervisor to gather more details on how to improve the work environment and employee retention.
“Look I don’t care if you’re sorry or not.” – Anger: 80%
In this example, that VERN detected a moderate amount of anger in their response. According to the business rules we established above, we don’t need to notify the legal team, but it could be worthwhile for HR to contact the employee’s supervisor to find out if there was a cause of the animosity.
Customer Service Rep Assistant
In this example, a junior customer service representative is online with a client having trouble with their financial account. V.E.R.N. is able to provide analysis in milliseconds, allowing for real time feedback on customer communication giving the rep useful information as they prepare to respond. One of the other advantages V.E.R.N. has is the ability to consider multiple signals and combine them to detect more complex emotions such as sarcasm.
# Anger and Humor Detection if anger_score > 75: # Strong anger signals detected if humor_score > 75: # Client may be saying something sarcastic # To Rep: Consider replying with a joke if appropriate else: # Client may be upset # To Rep: Consider elevating the to senior representative if behavior persists elif anger_score > 55: # Moderate anger signals detected # To Rep: The client may be indicating the source of their issue else: # Minimal anger signals detected # To Rep: Continue assisting as normal
As mentioned, this type of feedback can be sent directly to a representative as they’re actively in communication with a client. With the insights from V.E.R.N. a rep can more easily navigate the conversation and provide useful responses to resolve the issue. We can also see that by using a combination of anger and humor signals a more complex emotion may be communicated.
“By accident I paid a bill for my credit card using the wrong account and it was dinged for not having sufficient funds in my account!” – Anger: 60%, Humor: 33%
With this last example, we can see while there were traces of multiple emotions in the statement, the most dominant was Anger. Following the logic above, we can recommend to the representative to pay close attention to the sentence as it could point out the issue the customer is most upset about.
While these examples are not complete code, we hope they give you an idea on how V.E.R.N. can be used to implement business rules and decision making resources to make your system more efficient and provide a better experience for your users. If you’d like to see what kind of results V.E.R.N. can give you, check out our live demo and try it out!