In this presentation, we discuss our recent work of identifying ingredient substitution options using a knowledge graph of food and recipes. We will discuss the benefits provided by the linked data about ingredient nutrition and classification, as well as some key limitations that exist in the work so far. We further describe our plans for future work to incorporate various forms of context into our processes and how it fits in to the greater vision of the Health Empowerment by Analytics, Learning, and Semantics (HEALS) project.