Harnessing the power of cognitive and cloud technologies
In previous years, the AELTC had worked with IBM to develop a Social Command Centre—a hub that collects and analyses data from a variety of social media sources to give the tournament’s editorial team a clear view of how fans are reacting to the action. For the 2016 tournament, the team saw an opportunity to revolutionise the way it engages with social media by evolving the platform into a Cognitive Social Command Centre, leveraging IBM Watson™ technologies to reveal previously unseen levels of insight in real time.
Alexandra Willis comments: “First-mover advantage makes a significant difference in social media. The younger demographic tends to be much more ready to flick between different sources of information, so if you want to capture their attention—and keep it—you need to make sure you are continuously publishing the right new content in the right places at the right times.
“To create relevant and compelling content for social audiences, you need to keep your eye on the hot topics of conversation on each network at all times—and as the level of social media activity continues to increase, this is becoming almost impossible to achieve manually.”
To help deal with this big data challenge, IBM Watson Natural Language Classifier acts as an intelligent filter that uses machine learning techniques to “read” tweets and other social media posts and identify whether they are relevant to a given topic—in this case, tennis in general, and Wimbledon in particular. Instead of a naïve approach (for example, checking whether a given tweet contains a #wimbledon hashtag), the IBM Watson service is trained to understand unstructured text and learning from past experience to highlight messages that are relevant to the tournament.
To enrich this insight, the Cognitive Social Command Centre also uses the IBM Watson AlchemyLanguage service to pick out metadata from the text of the relevant social media posts—for example, which matches and players are being mentioned, and whether the sentiment appears to be positive or negative. Over the course of the tournament, metadata from over 17 million pieces of social content was aggregated and displayed in real-time dashboards in the Cognitive Social Command Center interface—giving the editorial team unprecedented levels of insight and inspiration about which stories were really resonating with viewers and fans.
“17 million posts sounds like a lot—but in fact, Watson significantly cut down the number of items we needed to analyse by intelligently classifying the messages that were actually relevant to us,” says Alexandra Willis.
“By separating the signal from the noise, the cognitive approach gives us a much more comprehensive view of what people say, think and feel about the tournament than we’ve ever had before—and in turn, helps us make quicker, more informed decisions about how to reach out to them.”
The solution also gave the AELTC a new appreciation of the lifecycle of the digital content it produces.
Alexandra Willis explains: “Typically, when we publish a new piece of content—for example, a video of a player winning a crucial rally—it follows a fairly predictable journey on social media. First it gets commented on and retweeted on Twitter and other ‘in the moment’ social channels. A few minutes or hours later, we see a peak in activity on networks such as Facebook, as people share the moment with their friends. And several hours later, often in the evenings, there’s a spike on YouTube and Vimeo, as people catch up or relive the excitement of the day’s play.”
She adds: “What’s interesting is when a piece of content doesn’t follow this pattern: perhaps it hits Facebook more quickly than expected, or it has an initial surge on Twitter but doesn’t really survive the transition to Facebook or YouTube. By seeing and understanding these patterns, we can gauge which kinds of stories are instantly exciting and memorable, and which are more of a slow-burner or don’t really resonate with viewers. This helps us tailor and adapt not only which content we are putting where, but also ensure we are adapting the structure of it effectively to each medium, and making better decisions about which stories to focus on.”
To give the team a head-start in generating exciting content in the first place, IBM also helped to extend the real-time notification system for breaking news that was launched with great success during the 2015 tournament.
“Last year we had real-time notification of a few key events—for example, when a player was about to reach a tournament milestone such as 1,000 aces or 5,000 winners at Wimbledon,” says Alexandra Willis. “This year we added more context, with real-time leaderboards that showed us how those achievements compared to other current players and all-time greats. Crucially, the system is smart enough to notify us before these events actually happen—giving us time to craft high-quality editorial content that really adds to the fans’ experience, and providing a vital advantage over our media competitors.”
Wimbledon also broadened the array of channels that it used to communicate these stories and engage with fans. New versions of its smartphone apps placed a greater focus on user-generated content, tying in with the social media strategy by making fan engagement more of a two-way conversation, and making the tournament feel more interactive for a younger generation who want to be able to create and share content, not just consume it.
In a world-first for tennis, IBM also helped the AELTC develop an app for Apple TV, which enabled viewers to tune into Live @ Wimbledon TV and radio broadcasts, plus live scores, photo galleries, and a huge range of video on demand content directly from their smart TVs.
Alexandra Willis comments: “Whichever way fans want to experience the tournament, whether it’s online, on their phone, on TV, or live at the venue itself, we want to give them the best possible experience—direct, uninterrupted, and authentic. The new apps are a key part of that strategy.”
The Cognitive Social Command Centre, as well as the responsive wimbledon.com website, mobile and TV apps and services such as SlamTracker®, are all powered by IBM hybrid cloud technology. IBM® Bluemix® provides a cloud platform for innovation, making it easy to build new apps and connect them with cognitive services and real-time data streams. Meanwhile, for more traditional web workloads, IBM SoftLayer® offers a massively scalable cloud platform that delivers a seamless online experience for millions of users and visitors. For extra capacity on demand, and to handle vital background tasks such as backups, IBM also hosts a private cloud infrastructure for the AELTC, running on IBM Power Systems™ servers.
“Cognitive and cloud technologies go together like strawberries and cream,” says Sam Seddon, IBM Client Executive for Wimbledon and RFU - Sports Marketing and Innovation. “We can easily combine IBM Watson services with real-time data feeds, and spin up new applications quickly without worrying about the infrastructure—helping us innovate faster than ever before.”
He adds: “Every year at Wimbledon is a challenge to deliver new capabilities based on leading-edge technologies, which have often never been tested before on such a large scale or in such a high-profile, high-stakes environment. Cloud services like Bluemix empower us to experiment and build prototypes quickly—but also to seamlessly scale up the best ideas into full deployment quickly as the tournament approaches.”
The hybrid cloud infrastructure is built to be able to meet the demands of millions of normal users over the course of the tournament—but it also needs to be protected against the small proportion of malicious users who seek to compromise its stability or security. For example, during the 2015 tournament, IBM observed a 500 percent increase in security attacks compared to the previous year. By early 2016, the scale of these attacks increased even more dramatically: April 2016 saw a 1,500 percent increase over the prior year. The IBM team knew that Wimbledon would need the best possible protection to reduce the risk of downtime or security breaches.
By using IBM Security Network Intrusion Prevention System and IBM Security SiteProtector™ System, and integrating all alerts into IBM Security QRadar® SIEM, the team was able to create policies, manage assets and analyse vast quantities of security data in a single place. This allowed the IBM team to pre-emptively detect threats and correlate groups of alerts into potential security events which could be managed proactively. As a result, the tournament passed with no significant disruption from security incidents.