Recommendation systems are crucial for personalizing content on video platforms. A main approach in video recommendation is item-based collaborative filtering, which predicts user interest by analyzing ratings of similar videos and aggregating these ratings using weighted influence factors. However, these systems face challenges in accurately computing video similarity and estimating influence between videos, as many existing methods rely solely on user experience data, such as numeric ratings. To address these limitations, we introduce a content-aware framework that integrates video ratings with video textual content. Our framework leverages advanced embedding techniques for feature extraction, a hybrid similarity measure combining ratings and content embeddings, and a dual loss function for influence estimation. Additionally, the system includes an explainability module to enhance user trust and engagement.