Consumption of fruits and vegetables has been linked to a reduced risk of cancer and other chronic diseases, but the molecular mechanisms supporting these connections remain largely unknown. A wealth of association data linking components of a plant-based diet, human genes, biological pathways, and phenotypes can be found in public databases and scientific literature. However, this massive amount of data is distributed across disparate sources, presenting a significant barrier to the investigation of the effects that plant-based diets impart on human health.This dissertation describes an integrated association network composed of existing curated and text-mined relationships which connect the agricultural and biomedical entities that define diet and disease. This research also describes HetERel, a meta path-based relevance ranking method for extracting highly relevant relationships between different types of entities in this network, such as a plant and the chemicals it produces. HetERel is tested on a network of chemical-disease association data for validation and performance. The method is then applied to the full-scale, integrated diet-disease network to discover distant, indirect links between plant-based chemicals and human phenotypes.The integrated diet-disease association network provides a foundational resource that connects plants and human health. Paired with the relevance search and prioritization method, HetERel, these methods empower researchers to generate hypotheses which elucidate the molecular mechanisms between plants and human disease.