Abstract: Drug application has greatly improved the quality of human life. The effectiveness of drug is a key factor in drug discovery process, and is determined by identifying drug-target interactions. However, it is a very expensive, time-consuming and challenging task to analyse and determine the compound-protein interactions through high-throughput screening experimental methods. The drug discovery research using computational methods are high efficiency and low cost, and it has been paid more and more attention. Compared with the wet-lab experiments, the computational prediction methods of compound-protein interactions can provide more accurate and safe potential candidate drug-target pairs for the subsequent biological experiment, and reduce the spending time and cost of biological experiments in drug discovery process. We review development of compound-protein interactions prediction, as well as the biomedical feature data, prediction algorithms, and technologies in the past two decades. This paper analyzes the problems faced in the research process such as high-dimensional sparsity of biomedical data and insufficient integration of multi-omics biomedical data, and it will provide valuable information for further research.