To develop the new drug and design strategies for diseases, it is required to understand the mechanism of herbal medicine. For this, it is important to construct networks of integrative relations among diseases, herbs, compounds and genes. Among these relations, it is necessary to understand the quantity information of compounds in herbs for high-resolution analysis about herbs have different effects to distinct phenotype. Most of this knowledge is available from the biomedical literature. In this study, we develop a machine-learning based text-mining model to extract herb-compound relations with quantity information from the biomedical literature.