Greetings SLR Community,
We excited to share the inner workings of our recently deployed Recommendation System at SLR. No marketing fluff, just the raw details.
Behind the Scenes: Tailoring Recommendations for You
Our Recommendation System serves as a guide, carefully curating a list of videos tailored to your preferences, based on your past actions. Currently on your Home page, its influence will soon extend to the Related Videos and Related Actors sections.
The Tech Unveiled: Streamlined for Performance
Handling a hefty amount of data, we honed our system for peak performance. The goal? Continuously reduce the items processed at each step, focusing on the most relevant ones.
Candidate Generation: Think of this as a collaborative filtering masterpiece. Our aim is to understand global similarity, irrespective of individual tastes. ML models help us discern how similar Scene A is to Scene B or User A is to User C, finding the perfect match for your queries. We look at factors like total view time, likes/dislikes, comments (sentiment and tonality), downloads, and more.
Ranking: With a curated list of candidates, we move to ranking. Here, your unique engagement with various aspects of our content takes center stage. A rich feature set, encompassing over 50 elements, helps us evaluate the importance of each feature, ensuring stability in the top factors.
What's Next? Your Questions and Insights!
Now that we've unveiled the mechanics of our Recommendation System, we want to hear from you. Do you have questions or insights to share? Have you discovered hidden gems through our recommendations? Share your thoughts, questions, or experiences below.
