Biomedical information is combined information between biological information and medical information. This biomedical information covers varying levels of units from the molecule level to the phenotype level. The introduction of omics data, including genomics, transcriptomics, proteomics, and metabolomics, made an integrated systematic analysis possible on biological phenomena. Hence, current biomedical information means multi-level relation data from numerous experiments. Electronic medical records data about patients’ information and other text data on biomedical literature are also included as biomedical data.Biomedical relations occur at specific conditions. According to recent research, each phase of a cell shows a different pattern of activated relations and every disease has various patterns of activated and deactivated relations. In addition, each gene expression shows context-specific patterns according to the kind of drugs used. Thus, understanding the variety of patterns under different conditions is essential to infer actual phenomena. Biomedical information needs an especially deep understanding of the dynamics of biomedical patterns under different conditions to derive effective treatments in various conditions. Previously, research aimed to discover biomedical patterns under specific conditions, but this research was not accurate enough to predict biomedical patterns under various conditions. A more accurate and automated methodology is needed to figure out context-specific patterns. Furthermore, an automated context interpretation method for given biomedical patterns should be developed.To ameliorate this situation, we suggested a methodology that discovers specific patterns in a specific context, as well as an automated context interpretation method for given patterns. We considered the context of disease and usages of drugs from primary biomedical information including experimental and reported text data to investigate relations between cont...