Jouska - Application de messagerie

This article presents a messaging app which is part of the bigger Jouska project - a speculative design project exploring the future of sentiment analysis.

Quantifying the way we talk

This project is a semi-working prototype of an app to quantify the way we talk. The app analyzes the valence of the user’s text messages. It compares the positivity of the messages sent and received, in order to see if there’s a loss of balance between the two people’s writing. It also looks at the personality of these contacts, to show which kind of personality is the most compatible with the user. Finally, it looks at the positivity of the messages depending on the weekdays/hours they were sent, in order, for example, to recommend people to avoid sending messages on certain days/hours, or on the contrary, to nudge them to do so.

Understanding its own conversation

The app is a normal messaging app in which you can see and send text messages.

Visualising emotions and personality

But it also contains a "Statistics" part, showing datavisualisations about the sentiment associated with each contact.

Statistics on the relationship with a certain contact. How is its personality compared to the user's one? How is its average positivity?

An emotional writing interface

More than showing data about the past discussions, the app also contains a writing interface that gives in real time an emotional feedback of the message currently written.

Illustration of how an emotional feedback can make people change the way they write. Left: "Hey what has happened to you yesterday? You looked sad..." Center: "Your message seems negative because of the word "sad" and the ellipsis" Right: "Hey Laura, how are you feeling today? :)"

Understanding the hidden mechanics underlying sentiment analysis

During the Brexit or after Donald Trump has been elected, people have started to claim that we now live in a post-factual era - an era in which public discourse is dealing with people’s emotions instead of objective facts. I think that the development of sentiment analysis algorithms has something to do with that phenomenon. Let’s think about Facebook new reaction buttons, that allow companies or political parties to have a more and more precise idea of our feelings.

For example, Facebook reaction buttons now allow companies to have a more subtle knowledge of our emotions.

If we prepare ourselves to enter a world that is using our emotions to nudge our impulse buys or to orient our votes, then it’s an act of resistance to create tools for making people understand the machinery behind sentiment analysis.

Developing an immunity at targeted solicitations

That's why the main purpose of the app is not to make people talk more efficiently.

The idea behind is, by allowing people to have a better understanding of the way they talk, the app trains them to read between the lines of one of the others.

This way, they can develop a form of immunity at targeted solicitations from advertisers, politicians, or noxious relations.