RNA plays important roles in cells through the interactions with proteins known as the RNA-binding proteins (RBP) and their binding motifs can provide crucial understanding of the post-transcriptional regulation of RNAs. How the RBPs correctly recognize the target RNA and why they bind specific positions is still far from clear. Artificial intelligence-based computational algorithms are widely acknowledged to be capable of speeding up this process. Although many automatic tools have been developed to predict the RNA-protein binding sites from the rapidly growing multi-resource data, e.g. from sequence, structure data etc, their different domain specific features and formats have posed significant computational challenges. One of current most difficulties is the cross-source shared common knowledge is usually at a higher abstraction level beyond the observed data, resulting in a low efficiency of direct integration of observed data in different domains. The other difficulty is how to interpret the prediction results. Existing approaches tend to terminate after outputting the potential discrete binding sites on the sequence, but how to assemble them into the meaningful binding motifs is a topic worth of further investigation. In viewing of these challenges, we proposed a deep learning-based framework (iDeep) by using a novel hybrid convolutional neural network and deep belief network to predict the RBP interaction sites and motifs on RNAs. This new protocol is featured by transforming the original observed data into a high-level abstraction feature space using multiple layers of learning blocks, where the shared representations across different domains are integrated. This bottom to up abstraction strategy is also very helpful to remove the noise effects in the observed data. To validate our iDeep method, we performed experiments on 31 large-scale CLIP-seq datasets, and our results show that by integrating multiple sources of data, the average AUC can be improved by 8% compared to the best single-source-based predictor; and through cross-domain knowledge integration at an abstraction level, it outperforms the state-of-the-art predictors by 6%. Besides the overall enhanced prediction performance, the convolutional neural network module embedded in iDeep is also able to automatically capture the interpretable RNA sequential binding motifs for RBPs. Large-scale experiments show that these mined binding motifs agree well with the experimentally verified results, suggesting iDeep is a promising approach in the real-world applications. The iDeep is available at https://github.com/xypan1232/iDeep.