Network pharmacology is an emerging approach for establishing the 'compounds-proteins/genes-diseases' network, aiding in drug target prediction through the interactions between the small molecules and the host microbes. In recent years, considerable information on the biochemical properties, location and biological properties of herbal phytochemicals and their interactions with the host through the microbiome have been accumulated and made available as repositories accessible to the general public. However, the scattered data from thousands of experiments, publications and other diverse sources make the storage, retrieval, in particular, data integration process, a cumbersome task. Additionally, new knowledge about the phytochemical properties and associated functions is constantly being discovered. Such a dynamic nature and multiple integrated relationships between the elements has made it difficult to manage the information with relational model of these databases because of its rigid schema. In the present study, we explore the effectiveness of using graph database (Neo4j) to store and effectively manage phytochemical data from 24 culinary herbs and genomic data for human microbiome. The HerbMicrobeDB (HbMDB) provides a graphically intuitive solution and simplifies the complex model, estimating the impact of herbs on microorganisms and in turn on the host.