A Survey on Food Ingredient Substitutions
Hyunwook Kim, Revathy Venkataramanan, Amit Sheth
TL;DR
This survey addresses AI-driven food ingredient substitution by synthesizing datasets, techniques, contextual knowledge, and safety considerations. It highlights datasets sourced from recipe sites, comments, and curated guidelines, and surveys methods that include rule-based, vector embeddings, and knowledge-graph approaches, with a growing emphasis on neuro-symbolic hybrids. The study identifies gaps such as lack of standardized benchmarks, limited integration of health labels and flavor context, and safety concerns requiring explainability and domain expert validation. It advocates for richer contextual knowledge, expert-curated data, and robust evaluation to enable reliable, personalized substitutions that support dietary needs and culinary creativity. A key open challenge is the development of a comprehensive neuro-symbolic framework that combines learning with semantic reasoning while ensuring safety and transparency for high-stakes dietary guidance.
Abstract
Diet plays a crucial role in managing chronic conditions and overall well-being. As people become more selective about their food choices, finding recipes that meet dietary needs is important. Ingredient substitution is key to adapting recipes for dietary restrictions, allergies, and availability constraints. However, identifying suitable substitutions is challenging as it requires analyzing the flavor, functionality, and health suitability of ingredients. With the advancement of AI, researchers have explored computational approaches to address ingredient substitution. This survey paper provides a comprehensive overview of the research in this area, focusing on five key aspects: (i) datasets and data sources used to support ingredient substitution research; (ii) techniques and approaches applied to solve substitution problems (iii) contextual information of ingredients considered, such as nutritional content, flavor, and pairing potential; (iv) applications for which substitution models have been developed, including dietary restrictions, constraints, and missing ingredients; (v) safety and transparency of substitution models, focusing on user trust and health concerns. The survey also highlights promising directions for future research, such as integrating neuro-symbolic techniques for deep learning and utilizing knowledge graphs for improved reasoning, aiming to guide advancements in food computation and ingredient substitution.