In today’s world where digital services are everywhere, the service providers main focus is user satisfaction. To maximize user satisfaction, it is crucial to provide relevant and authentic recommendations. Our model Deep Hierarchical attention mechanism for Cross Domain Recommendation (DeepHM-CDR) provides a resilient approach, which not only extracts the relevant services but also authentic ones. Our model merges a deep authentic review detection model i.e. DeepHM which uses the hierarchical attention mechanism to segregate the fraudulent reviews from the real ones. The second phase of the model Bidirectional Transfer Mechanism for Cross-Domain Recommendation i.e. DBTM optimizes the recommendation process in alignment with the authentic reviews. Thus both the models work together to enhance user satisfaction by providing creditable and suitable recommendations. Moreover, because of simultaneously considering information from two domains, it solves the problem of cold start users to a good extent. Our model is evaluated for two real-world benchmark datasets i.e. Douban dataset and the Amazon 5-crore dataset for both movie and book domains and the results clearly show that it outperforms the state-of-the-art approaches.