Amazon Prime Video Enhances Sports Search with Amazon OpenSearch Service

Passionate sports fans are always looking for an easy way to find their favorite teams and games. That’s why Amazon Prime Video is constantly working to improve its search experience for sports content. With a huge selection of live and on-demand sports options, Prime Video knows that a smooth search process is key to keeping fans engaged and happy.

The search bar on Prime Video is one of the most popular features, allowing users to quickly find the content they love. By refining their search capabilities, Prime Video can offer personalized recommendations that keep viewers coming back for more. This attention to detail not only enhances the customer experience but also builds trust and loyalty among sports fans, driving long-term success for the platform.

One of the challenges Prime Video faced was adapting their search system, originally designed for movies and TV shows, to include sports content. Unlike movies, sports events are more time-sensitive and seasonal. This meant that the existing search system wasn’t effectively surfacing live sports events to users. By revamping their search capabilities with Amazon OpenSearch Service in 2024, Prime Video was able to create a more intelligent and intuitive search system tailored specifically for sports fans.

The new search functionality uses semantic search and binary search relevance classification to ensure that users find relevant sports events quickly. Semantic search goes beyond simple keyword matching, using vector embeddings to understand the meaning behind words, phrases, and sentences. This allows the system to recommend sports events based on similarity in meaning, even if the words themselves don’t match exactly.

Vector embeddings are created for each sports event in the Prime Video catalog, capturing important details such as team names, leagues, and event specifics. When a user searches for something related to sports, their query is also converted into a vector to find the most relevant matches. A machine learning model filters out irrelevant results, leaving behind a refined list of the most pertinent sports events for the user.

By combining vector semantic search with relevance classification, Prime Video ensures that users can easily access the live, upcoming, and recently ended games they’re most interested in. This sophisticated search system not only improves the user experience but also keeps sports fans engaged and coming back for more.

To make all of this possible, Prime Video used AWS services like Amazon SageMaker and OpenSearch Service to host their custom text embedding model and KNN index. This infrastructure allows Prime Video to handle thousands of search requests per second, ensuring a seamless experience for sports fans around the world.