In the evolving landscape of Web 3.0, characterized by enhanced cohesiveness and data density, there is a distinct lack of comprehensive frameworks for video recommendations, especially in niche domains like mobile journalism. This paper introduces a robust strategic framework aimed at addressing this gap. Focusing specifically on the unique domain of mobile journalism, the framework employs a sophisticated learning strategy that combines Convolutional Neural Networks (CNNs) and Decision Trees, effectively balancing computational load and optimizing video recommendations. A key feature of this framework is its ability to synthesize data from standardized knowledge repositories and knowledge graphs, enriching the recommendation process with auxiliary knowledge. To enhance semantic reasoning and recommendation quality, the framework leverages metrics like the Jiang Conrath similarity index, Simpsons Diversity Index, and Generalized Entropy Index. Operating as an incremental model, it captures knowledge from both query and dataset ends, ensuring that the resulting recommendations for mobile journalism videos are top tier, tailored to the specific preferences and needs of users in this specialized domain. An overall precision of 95.09%, F-measure of 96.47% and the lowest value of FDR, which is at a value of 0.05, has been achieved by the proposed framework.