Purpose/Significance To discuss the application of network representation learning and link prediction in mining potential medical entity relationships, and to provide a novel perspective for medical knowledge discovery. Method/Process The literature abstracts are obtained from PubMed database, disease drug treatment information within these abstracts is identified using the subject-action-object (SAO) semantic mining scheme. Drug entities and disease entities are extracted, and a “drug-disease” bipartite network is constructed. The network structure and node characteristics are analyzed comprehensively using methods of social network analysis, network representation learning and machine learning, potential connections between medical entities are explored. Result/Conclusion The bipartite network representation learning method can reveal the association knowledge between drugs and diseases, identify key medications for treating diseases, and provide feasible treatment options.