Google Professional Machine Learning Engineer
Get started today
Ultimate access to all questions.
Comments
Loading comments...
Ultimate access to all questions.
You are leading the development of a new video streaming platform and have been tasked with designing a recommendation system to suggest the next video for users. The platform's video catalog includes metadata such as release date, country, and content type, but there is no historical user event data available initially. The company aims to launch the platform quickly while ensuring scalability and the ability to improve recommendations over time. Considering these constraints, how should the initial version of the product's recommendation system be developed? (Choose two correct options if E is available, otherwise choose one.)
A
Begin without machine learning. Use simple heuristics based on content metadata to recommend similar videos to users, and start gathering user event data for future recommender model development.
B
Start with machine learning. Create video embeddings by training an autoencoder on the content metadata using TensorFlow. Group videos by embedding similarity and recommend from the same cluster.
C
Launch without machine learning. Present videos to users in alphabetical order, and begin collecting user event data for future model training.
D
Implement machine learning from the start. Use a public dataset like MovieLens to train a model with Recommendations AI, then apply this model to your data.
E
Combine both heuristic and machine learning approaches. Use content-based filtering initially and gradually incorporate collaborative filtering as user event data becomes available.