The exponential growth of scientific publications challenges researchers to efficiently extract meaningful knowledge. In the context of food, diseases, and drug interactions, this project aims to construct a comprehensive knowledge graph using scientific literature and predict novel relationships within it. For this purpose, we will review existing literature on food, diseases, drug interactions, and knowledge graph (KG) construction. We will identify key publications and datasets crucial to the project. Moreover, we will utilize natural language processing (NLP) techniques for extracting relevant information from scientific publications. Next, we will integrate curated datasets on food, diseases, and drug interactions into the knowledge graph. In addition, algorithms to structure information into a knowledge graph will be developed, defining entities (e.g., foods and its constituents, drugs) and relationships (e.g., causes, interactions). Semantic enrichment techniques to enhance the knowledge graph will be implemented. Next, we will apply semantic analysis tools such as knowledge graph embeddings to identify patterns and associations within the knowledge graph. By utilizing machine learning algorithms (link prediction), we will predict potential new relationships within the knowledge graph. Further, we will develop an interactive and user-friendly interface for researchers and healthcare practitioners to explore the knowledge graph. The accuracy and reliability of predicted relationships will be validated by comparing them with known relationships from independent datasets. In addition, feedback from domain experts will be collected and to help researchers to refine and enhance the knowledge graph and prediction models. The results will provide a valuable resource for researchers, healthcare professionals, and policymakers to make informed decisions.

bilateral ARIS - ZDA