Published Paper Details:

Food Recommender Systems (FRS) has emerged as an important sub area of Computer Science and Engineering (CSE) that is significantly different from other traditional areas of recommendation such as e-commerce or media streaming. This systematic review aims to do an architectural analysis of intelligent FRS using Artificial Intelligence (AI) and Machine Learning (ML). We summarize 21 basic peer-reviewed studies and examine how these emergent systems cope with high dimensional and non-linear constraints like user health states, complex multi-component ingredients, temporal consumption rules, and multimodal streams of features. These methods fall into four structural paradigm groups: Traditional Matrix Methods, Sequence-based Neural Nets, Heterogeneous Information Networks (HIN), and Advanced Optimization Frameworks. Last, open challenges for sparsity training, cold-start issue and unified offline metrics for algorithmic precision nutrition are elaborated which will serve as future research roadmap.