AI Steering the Narrative of Roland-Garros

Hard-fought matches at Roland-Garros can last as long as six hours. Now, distilling the essence of such epic battles on the clay courts in Paris can take less than two minutes.

New artificial intelligence (AI) tools allow the Roland-Garros editorial team to view data analysis mid-game. Complete summaries — charting the critical breaks, errors, and holds that led to a win — are ready moments after the match ends. The results have helped shape more than two dozen articles on the Roland-Garros website this fall.

The new AI-assisted journalism portal is part of the French Tennis Federation’s technology initiatives, designed to entertain and inform fans. Roland-Garros has partnered with tech company Infosys to expand the Grand Slam’s use of AI and data analytics.

“When you’re in the stadium, you’re drawn to the emotion,” said Craig O’Shannessy, one of tennis’s leading strategists. “Remotely, the focus shifts to a new statistical context — that is, stories made with data.”

Originally planned for late May and early June, Roland-Garros was postponed due to COVID-19. The tournament, also known as the French Open, was rescheduled for Sept. 21-Oct. 11. The pandemic has severely restricted crowds — 1,000 fans per day compared to 35,000 in a normal year. However, millions are following the tournament online and in the media, often expecting updates instantly.

The speed of AI

The AI-assisted journalism system at Roland-Garros provides content creators with deep analysis almost instantly, speeding up an already fast online metabolism. That allows writers and producers working for the Roland-Garros website to quickly deliver rich content, including articles and graphics that capture the nuances buried deep in the matches.

Roland-Garros website staff can log on to the Infosys portal and review the factors that led to each victory, including serves, returns, rallies, and net points. The writers and producers also see the players’ most telling performance statistics. Every match is covered by the AI portal, although some features are not available on courts that don’t have the full complement of technology.

The AI system generates strength, weakness, opportunity, and threat analysis videos that offer another way to understand how matches were played and won. Producers can also choose a particularly important data point, and the system can generate a graphic illustrating that insight.

Use is not mandatory, but content creators have taken extensive advantage of the tool. The system also produces AI-generated video highlights of the match’s most pivotal moments, so writers can review those instances and draw their own conclusions.

“If they have any lingering doubts about what we picked as the most important moments, they can see for themselves and decide whether they are convinced,” said Raghavan Subramanian, head of tennis platforms for Infosys.

The tool’s ultimate goal is to cut through the thousands of events and lead the editorial staff toward the most meaningful ones. It’s easy to recall the exciting or dramatic moments. But what about the quieter points or the surprising patterns that steered the match toward its conclusion?

Next steps for AI

 This Infosys AI tool is currently available only to the Roland-Garros editorial staff. But there is an expectation that its use could be expanded. “What we do for sports can be easily extrapolated for any other industry and data,” Subramanian said.

In the short term, Infosys plans to improve the algorithm’s use of natural language processing. “Today we are capable of looking at a set and giving a one-line insight that says what exactly transpired,” Subramanian said. “It would say something like, ‘[Rafael] Nadal came from behind to win the set against [Dominic] Thiem by making fewer unforced errors.’”

Subramanian said the next step is to create a sentence — combining fact and cause — for every point in the match. Instead of saying the shot was long, the system would be able to say the player was attempting a deep sliced return that was long. “There, you are giving causality,” he said.

When that is available, the AI can be pointed toward the problem of turning those individual sentences into paragraphs, passages, and entire narratives. “Then we would have a nice little milestone,” Subramanian said.

Even without those advances, the Roland-Garros editorial staff will still have a shortcut to help with the most pressure-filled moments of their year.

Behind the #15Love Experience: This blog series celebrates the spirit of collaboration and innovation that went into shaping the Roland-Garros 2020 digital landscape. Follow Infosys on LinkedIn to hear stories of passion and purpose from the teams at Infosys and French Tennis Federation (FFT), as they talk about the next normal in experiences, partnerships, technology and branding.