Design an extension to the collaborative filtering model that can leverage additional information about users and items to mitigate the cold-start problem. Explain the tradeoff between using more features versus increasing computational complexity.

Introduce user profiling features (e.g., demographics, interests) and item features (e.g., categories, descriptions) into the recommendation system. This allows for similarity evaluation even when new users or items lack rating history, but may increase computational complexity due to additional feature processing and integration with existing models.