Every day millions of people tune in to watch their favourite streamers. These streamers often have very passionate followers, upon whom they can have a large impact. It is therefore important for streamers to understand how their words and actions impact their viewers.
However, many of these streams can have thousands of concurrent viewers at any given time. Moreover, viewers usually communicate with the streamer through text chat, which lacks much of the feedback that exists in physical contexts with a live audience. This can make it hard for the streamer to keep up with what goes on in their chats, especially while they focus on other activities such as playing a game.
Audiment helps them do so by presenting the streamer with live sentiment analysis of their chat. This can allow the streamer to see how the chat is reacting to their stream at a glance.
Audiment enriches viewer-streamer interaction using sentiment analyses. During the stream, the streamer gets live information about the current sentiment in their chats. After the stream ends they also get a report where they can analyse the moments that caused a large shift in viewer sentiment.
Auditar is an avatar that can express a number of emotions to reflect the current mood of the chat. This allows the streamer to gauge the current sentiment on the chat at a glance.
The Sentiment Graph captures how the sentiment within the chat has evolved throughout the stream. The streamer can use this to catch up with sentiment developments in between games.
The Impact Cloud displays the phrases that the streamer has said during the stream that had the biggest effect on sentiment. Moreover, colours are used to communicate if the phrase had a positive or a negative effect on sentiment. This is done by taking the stream's audio and converting it to text. Then each phrase is rated based on the change in sentiment over a short interval after it was said. This allows the streamer to link large changes in sentiment to the thing they said.
The Post-Stream Report offers the streamer an in-depth sentiment analysis of their streams that they can view while offline. The streamer can use this to develop a deeper understanding of the impact of their actions and words on viewer sentiment.
Future versions could allow streamers to focus on specific types of viewer sentiment. For example, Audiment could allow the streamer to flag spikes in negative sentiment, even in cases where it was drowned out by positive sentiment. This could allow the streamer to notice when a joke that is generally well-received, may have offended some parts of their audience.
The prototype is accessible here
How to use: Click play on the video and the prototype will simulate the demo (Please make sure that you use Google Chrome and that you have a fast internet connection otherwise the demo will break due to buffering.)