Welcome to the third in our series of posts highlighting the top 10 use cases driving innovation at the IBM Machine Learning Hub, where we're working hand-in-hand with clients to see into the future of machine learning.
The goal of the series is to look beyond the usual set of machine learning use cases that come to mind when considering a particular sector. Part 1 explored novel use cases in the realm of government that are driving improvements to regional and municipal agencies. Part 2 looked at the influence of ML on mental health, whole-life health, and reducing re-admission rates.
Here in Part 3, we'll take a look at something a little lighter: the world of media and entertainment. Our first association with machine learning in media and entertainment might be the recommendation engines that offer personalized suggestions for books and films, but IBM clients are turning to ML for help with everything from copyright tracking to optimizing data warehouses to letting tennis fans monitor and analyze tournament statistics in real time.
1. Helping Artists to Get Their Due
With the explosion of digital distribution, authors, composers, and other artists are struggling to reap the rewards for the creative work they generate. The more popular a song, story, or film, the more likely it is to be pirated and shared with no thought for compensating the artists who created it. For independent artists, even a small increase in royalty payments can spell the difference between paying the bills by doing what they love - or turning away from their life's work to make a living in other ways.
Consider the European nonprofit organization whose core mission is collecting royalty payments for songwriters, composers, and music publishers. Owned collectively by its 150,000 members, it represents over 100 million works around the world. In 2016, the organization tracked around 982 billion download and streaming transactions, an incredible volume of data that they knew contained the unique identifiers within metadata that the organization could use to seek compensation from companies like Apple, YouTube, Spotify, and Facebook.
In alliance with IBM, the organization developed a cognitive copyright tracking platform that looks not only at the metadata but even deeper into the creative content itself, using natural language processing (NLP) and pattern recognition to identify an even greater share of the consumed content. That's expected to deliver a 15 percent increase in royalties to its members — while also helping them understand trends in how, when, where, and by whom their work is being consumed.
2. ML and a Smarter Warehouse for Media
Let's turn from the creators of entertainment content to the distributors. The media giants making content available digitally face a constant challenge to keep their data warehouses up to date with the shows, movies, songs, and games that their users are demanding. Staying competitive means getting the right content up and available fast.
But speed and capacity aren't enough. An IBM client in America needed to go further and generate analysis in real time of which content was in demand. With IBM's help, the analytics they generated could indicate not just what had happened and why but could go a step further and offered predictions about which new content will strike a chord with users — by history, geography, time of day, and more. That's allowing them to allocate storage and streaming resources in advance to give every user a seamless experience.
3. Advantage Machine Learning
To true fans, nothing feels better than getting close to the game, and for plenty of tennis fans that means the ability to study and interrogate the raw statistics of scores, aces, faults, serve speed, player position, and more. Yet it's one thing to do that after a match once the details have been slowly gathered and compiled. It's something else completely to be able to dive into that data in real-time as the rackets are still swinging.
That's exactly what happened at a premier Grand Slam tournament when IBM teamed up with one of the largest tennis organizations in the world — an organization with more than 17 geographical divisions, 750,000 members, and 7,000 organizational members. Using a combination of predictive analytics, cloud, mobile and social technologies, the team was able to integrate the live match data with the historical player, match, and tournament data. The results powered a set of custom scoreboards and interactive displays fully available to reporters on site and fans online.
Where does machine learning come in? Based on the analysis, the software could identify three key strategies likely to affect the scoring and dynamics of any given match. And as a match proceeded, the reporters and fans could track each player's progress against the predictions in real time.
We all know that contemporary media and entertainment is increasingly digital and driven by data. For those tasked with keeping audiences happy and engaged, there's no substitute for the analysis and predictions made possible by machine learning algorithms that are tuned to anticipate their desires.
But just as important is what happens behind the scenes. That's where IBM is helping to compensate the creators of content while encouraging fans to engage as creators themselves — by turning raw statistics into new forms of interactive entertainment. Across use cases, IBM sees its media and entertainment clients using machine learning not only to improve user experiences but to promote fairness, foster independence, and fuel passion. It's incredible work — but just the beginning of what ML will make possible.