At any time of the day, anywhere in the world, more than one-half of the planet’s population has the ability to pick up their smartphones and transform themselves into actors, authors and artists. They can create and share in seconds.
This shift in the creative power has led to the rise of a new form of celebrity–the social influencer.
YouTube, Instagram, Facebook and Snapchat creators are now more popular than mainstream celebrities.
This is a remarkable shift, which means that modern-day fans have the power to choose who is famous to them. Who will they choose and why? And can machine learning successfully identify the next big social celebrities?
In machine learning, computers apply statistical learning techniques to automatically identify patterns in data. These techniques, once properly refined, can be used to make highly accurate predictions. The process is achieved by creating models through training data that repeatedly incorporate “if-then” statements, or “forks.” Based on existing data, the model continues to apply such statements to new data in order to reach a conclusion that is filtered down until satisfactorily accurate. Thus far, machine learning has proven to predict health-related complications, credit risk and even suicide.
As for applying machine learning to social media, it is a complex but fascinating proposition, because users interact with one another and with the content they both create and consume. This gives researchers access to user interactions that are much richer.
Applied to decision trees and logistic regression, this information can result in many predictive features such as insight on audience reactions (likes and dislikes), as well as the ability to predict the level of social popularity simply from the social media status updates and similarities of the users.
Having said this, thus far, the primary use of the predictive function of machine learning has been for aligning content of interest, suggesting connections to other users and providing a deeper level of engagement with brands wishing to use social media in their marketing mixes.
As it stands today, social celebrities are passionate, connected individuals who have gained a massive following in their area of interest. Their importance seems to be a matter of consensus, whether it be U.S. presidential candidates relying on endorsements from social celebrities or millennials who overwhelmingly declare the internet as the “one thing they could not live without.”
Nevertheless, the application of machine learning to the metrics of their fame is generally unchartered. In fact, it is really the first time that we may actually be able to predict not only fame, but degrees of fame, as well as the road to ascension.
The interest of such predictions is twofold; first, brands seeking to expand their reach through social media will be able to align themselves with social influencers according to future spikes in popularity. More important, the pool of followers that contribute to this fame is of particular importance, as it will allow for seamless brand integration and demographic alignment.
Second, for the influencers themselves, it will provide them with deeper insights into their fame and, thus, their value in promotional partnerships.
Although the idea of predicting fame seems somewhat far-fetched, it is largely based on a solid mix of well-established factors, temporal variations and T-density, with an application of regression algorithms.
Simply put, similar to predicting any other component of social media user behavior, continued research in the field will undoubtedly result in a “secret sauce” that will allow us to determine who the stars of tomorrow will be.
The idea of a future where social celebrities can be anticipated has the potential to disrupt many industries but will also lead to a more democratic process where celebrities will accurately represent the fans who look up to them.
Jon Bucci is the CEO of influencer marketing network Veri.