Revolutionizing Sports Search: FOX Sports' AI-Powered Fan Experience

From Lagging to Leading: Addressing the Need for Change

FOX Sports has always been at the forefront of broadcast innovation, from pioneering high-definition broadcasts to exploring virtual reality. As the digital world evolved, we recognized the need to enhance our online search, aiming for a seamless experience in the face of ever-increasing online content.

The old search simply wasn't cutting it for our fans. Imagine searching "Lionel" and sifting through countless results before finding soccer superstar Lionel Messi. Or trying "christian pulis" only to be met with unrelated "Christians" and "Pulises" instead of US Men's National Team star Christian Pulisic. Fans craved speed and accuracy, but our fragmented system forced them to meticulously craft queries, leading to frustration and a disjointed experience.

Redefining Search: The Power of AI and Real-Time Data

Partnering with Databricks, we embraced the potential of AI-powered semantic search. This goes beyond simple keyword matching, understanding the intent behind a search. It accounts for synonyms, misspellings, and complex relationships between topics, delivering intuitive results that align with user expectations.

Real-time data ingestion pipelines, powered by Spark Structured Streaming and Databricks Workflows, constantly update our system with fresh entity data (athletes, teams, leagues), articles, videos, and third-party content. This dynamic approach keeps our search relevant in the fast-paced world of sports, reflecting roster changes, breaking news, and shifting fan interests.

For example, searching "Washington" might prioritize the Nationals during baseball season and the Commanders during football season. However, if the Commanders make a major signing during baseball season, our intelligent search engine will recognize the surge in fan interest and prioritize that information.

Elevating the Fan Experience: Speed, Context, and Discovery

Our vision for search is driven by three key goals: enabling natural, contextual searches; integrating semantic search for richer content discovery; and delivering a real-time, low-latency experience. We want search to adapt to the fan, not the other way around.

With Mosaic AI Model Serving and Vector Search, we've built a system that prioritizes exact matches for entities while offering time-weighted semantic search for content. Now, a search for "Cowboys offseason updates" surfaces a rich tapestry of information, including Dak Prescott, injury updates, and expert analysis from NFL on FOX.

We've streamlined navigation by directing users to specific subpages for targeted queries. Searching for "MLB schedule" takes you directly to the schedule page, eliminating unnecessary clicks. Our popular searches feature, powered by real-time user interaction data, highlights trending topics before a user even begins typing.

A New Era of Sports News: Agile, Efficient, and Engaging

The results have been remarkable. Our new endpoint handles hundreds of thousands of requests daily, with significant spikes during weekends and live events. A caching layer optimizes performance, and our popular searches feature alone accounts for over 25% of all searches, demonstrating the value of real-time data intelligence.

A Databricks Lakehouse App visually showcases the improvements, securing stakeholder confidence in this transformative upgrade. We've not only improved the fan experience but also reinforced FOX Sports' commitment to innovation.

Through Databricks, we've built a robust data and AI foundation, blending real-time performance, cutting-edge technology, and dynamic search adaptability. At FOX Sports, we're not just keeping pace with the evolving demands of sports fans – we're staying a step ahead.