Sentiment Analysis

Aware comes with a feature called Spotlight that uses natural language processing to determine the sentiment of a user’s message, as well as the overall sentiment within teams in the workplace. The first picture below shows what would be displayed if an Aware user was setting up preferences for what degrees of sentiment they wish to monitor.

Interestingly, designing for this wasn’t so much a visual design issue as it was a copywriting problem. I was given the task of meeting with the head data science engineer in order to determine the proper diction for a typical statistical analysis; While it may seem as though I might have been able to simply ask a few questions and move forward with the design, it turned out to be a lesson in statistical reporting standards. In order to ethically represent the data most accurately, went into such specifics as to change out modifiers such as, “Very Positive,” and, “Very Negative,” for the more statistically appropriate, “Strongly Positive,” and, “Strongly Negative.”

I bring up this example because it demonstrates the imperative nature of relying on the expertise of our peers. Without his guidance, I wouldn’t have been able to appropriately represent the data in my design, leading to an inferior experience with our users.

The second image simply displays the hover state prototype, showing the feedback responsiveness of the product.

The third and final image displays another feedback feature in the form of a message appearing when all of the radio boxes are checked; The message warns the user that scanning for all sentiment options might produce an overwhelming amount of events and false positives, rendering the feature useless by casting too wide of a net. This added bit of guidance should help the user gather more valuable data and feel more satisfied with our product.