To design new preventive and therapeutic strategies for diseases, it is necessary to understand relationships between diseases and drugs in a variety of perspectives such as the clinical-centric and the molecular-centric approach. In such approaches, the information about clinical factors and gene expressions are used for intermediates that link diseases and drugs. In this regard, it is important to collect the accurate and specific information about relationships between biomedical entities (disease, clinical factor, gene expression, and drug) in order to precisely interpret disease-drug relationships. Most of this knowledge is available via the biomedicalliterature. In this thesis, we introduce machine-learning based text-mining frameworks to extract relationships between biomedical entities from the scientific publications.