Microsoft today announced a new feature for Bing.com where they will be posting the predictions of the events based on their ability to process trillions of signals to reflect what is happening in the real world.
For instance, during the presidential elections, Bing experiences a surge of positive and negative queries about the presidential candidates from different parts of the U.S. Taking this model even further, teams within Bing have been experimenting with useful ways that we can harness the power of Bing to model outcomes of events. This could be anything from an election or sporting event to whether an actor will win the People’s Choice Award.
Bing team is bringing these insights directly to our search results pages. Based on a variety of different signals including search queries and social input from Facebook and Twitter, they will be unveiling an experiment they have built to give you our prediction of the outcome of a given event.
How does it work? To trigger the feature, simply search for The Voiceor a current contestant on the program. Based on your search, we will display a carousel that provides our estimation of who is on top, who is in danger and who is likely to be eliminated. Here are a few sample queries:
How do we make the predictions for The Voice?
The central idea behind the direct approach is that winners and losers correspond to popularity. In broad strokes, we define popularity as the frequency and sentiment of searches combined with social signals and keywords. Placing these signals into our model, we can predict the outcome an event with high confidence. For The Voice, our model is tuned to account for biases, such as regional preferences, and other measurable and observable trends. We have also learned how to combine the prior popularity of a contestant with the contribution due to his or her performance. Although we might believe that the outcome of this week’s The Voicecomes down to how well someone belts out a tune, our data indicates that many people have ‘favorites’ regardless of individual week-to-week performances.
Read more about it here.