At Virtual U.S. Open, IBM Serves Up New Tech Tools for Fan Engagement

When it comes to tech + sports, the U.S. Open has already won a few majors. Since 2017, each match’s highlight reel has been determined by IBM-created ML algorithms that analyze crowd reactions.

Enter COVID-19: The virtual U.S. Open has no crowd reactions to analyze. What it does have: high expectations for at-home viewership.

Taking a swing on NLP

How do you keep the attention of homebound viewers used to sipping Honey Deuces in stadium seats? The USTA partnered with IBM to develop two new natural language processing tools, launched last week:

• Match Insights: Fans browsing the site ahead of an upcoming match can see algorithmically selected facts, which highlight the “largest percentile differences” between two players. That data is then translated into ~human words~.
• Open Questions: This feature analyzes millions of online sources to gather pros and cons—and rate argument quality—for tennis’ longstanding debates. Viewers can weigh in, and their opinions will become data points that are algorithmically summarized on the site at 5 p.m.

Big picture: Watson isn’t the only IBM headliner at play—there’s also the evidence-based reasoning involved in Project Debater and DeepQA (Jeopardy!). By pouring money into this tech, IBM’s betting on the future of a still-emerging industry: machine writing.