CHAPTER TWO

LITERATURE REVIEW
2.1 Introduction
The integration of nutritional science with Artificial Intelligence (AI) has evolved significantly
over the years. Traditionally, dietary recommendations were largely generalized, offering
uniform guidance without considering individual differences. However, recent advancements in
research emphasize a more personalized, “person-centric” approach to nutrition. This approach
recognizes that effective dietary recommendations must account for individual characteristics
such as body measurements, medical conditions, lifestyle patterns, and personal preferences
(Wang et al., 2025).
With the rapid growth of intelligent systems, AI-driven dietary recommendation tools have
transitioned from simple tracking applications to sophisticated decision-support systems. These
systems are capable of analyzing complex datasets and generating tailored nutritional advice.
This chapter reviews the theoretical foundations of expert systems, compares rule-based and
machine learning approaches, and examines existing diet recommendation systems. It also
identifies key research gaps that justify the development of the proposed system.

2.2 Theoretical Review

2.2.1 Knowledge Engineering and Expert System Foundations
The development of a Nutrition and Diet Recommendation System is grounded in the principles
of Knowledge Engineering. Expert systems, as defined by Hadi (2024), are computer programs
designed to replicate the decision-making ability of human experts by applying logical reasoning
to structured knowledge.
A key characteristic of expert systems is the separation between the knowledge base and the
reasoning mechanism. This distinction allows for easier updates and better system transparency
(Durkin, 2024). The architecture of an expert system is typically composed of three main
components:
The Knowledge Base:
The knowledge base contains domain-specific facts and rules derived from experts. These rules
are often expressed as IF-THEN statements, known as production rules. According to Lucas and
van der Gaag (1991), these rules form the backbone of expert systems, enabling them to simulate
expert reasoning. For example, a rule may state: IF a patient has high blood sugar, THEN
recommend a low-carbohydrate diet.
The Inference Engine:
The inference engine is responsible for applying logical reasoning to the knowledge base. In this
study, forward chaining is employed as the reasoning technique. Forward chaining is a data-
driven approach that starts with known facts and applies relevant rules to reach conclusions.
Ikawati and Ramadhan (2026) highlight that this method is particularly effective in systems that
rely on real-time user input, such as dietary recommendation platforms.
User Interface and Explanation Facility:
The user interface allows users to interact with the system, while the explanation facility
provides justification for the system’s recommendations. This feature is essential in healthcare
applications, as it enhances user trust and ensures transparency in decision-making (Hadi, 2024).

2.2.2 Rule-Based Systems versus Machine Learning Approaches

In modern AI applications, both rule-based systems and machine learning techniques are widely
used. Machine learning models are capable of analyzing large datasets and identifying patterns,
making them suitable for predictive tasks. However, these models often lack interpretability and
are commonly referred to as “black-box” systems (Butti & Gangappa, 2020).
In contrast, rule-based systems offer a higher level of transparency and explainability. Each
decision made by the system can be traced back to a specific rule, making it easier for healthcare
professionals to validate recommendations. Stefanidis et al. (2022) emphasize that rule-based
systems are particularly valuable in healthcare because they adhere strictly to established clinical
guidelines.
Although machine learning enhances adaptability and personalization, its lack of explainability
can limit its application in sensitive domains. As a result, many researchers advocate for hybrid
approaches that combine the strengths of both methods (Lambay & Mohideen, 2022).

2.3 Review of Related Systems

2.3.1 Machine Learning and Deep Learning-Based Systems
Recent studies have explored the use of machine learning and deep learning techniques to
improve dietary recommendation systems. Kumari et al. (2024) developed a personalized diet
recommendation system using the Nearest Neighbor Algorithm and Cosine Similarity. Their
approach matches user profiles with food databases to generate recommendations that align with
individual nutritional needs.
Similarly, Ahmed et al. (2025) proposed a Diabetes-Focused Recommender System (DFRS)
based on Recurrent Neural Networks (RNNs). Their findings demonstrate that deep learning
models are capable of handling complex and high-dimensional health data, enabling more
accurate and dynamic recommendations for patients with chronic conditions.
In another study, Fisichella et al. (2025) introduced Nutri-Graph, which utilizes Graph Neural
Networks to enhance food recommendation accuracy by modeling relationships between
nutrients and food items.
2.3.2 Generative AI and Large Language Models
The emergence of Generative AI has further advanced the capabilities of dietary
recommendation systems. Papastratis et al. (2024) proposed a system that integrates deep
generative models with large language models such as ChatGPT. This approach enables the
generation of diverse meal options tailored to user preferences.
However, generative models must be carefully controlled to ensure nutritional accuracy. Without
proper constraints, they may produce recommendations that do not meet clinical standards. To
address this limitation, researchers recommend combining generative models with expert-defined
rules and optimization techniques (Papastratis et al., 2024).
2.3.3 Hybrid Systems and Cultural Adaptation
Hybrid recommendation systems combine rule-based logic with machine learning techniques to
achieve both accuracy and flexibility. Lambay and Mohideen (2022) highlight that such systems
can leverage data-driven insights while maintaining adherence to expert knowledge.
In addition, cultural and geographical factors play a significant role in dietary recommendations.
Vandeputte et al. (2023) introduced the concept of “acceptable food substitutions,” where
systems suggest gradual dietary changes that align with user preferences. This approach
improves user acceptance and long-term adherence.
Kalpakoglou et al. (2025) further emphasize the importance of integrating local food
composition databases to ensure that recommendations are relevant to specific regions. This is
particularly important in developing countries, where food availability and cultural preferences
vary significantly.
2.4 Research Gaps
Despite the progress made in AI-based dietary recommendation systems, several limitations
remain.
Firstly, there is an oversight gap, as many systems are designed for individual users without
incorporating professional supervision. Garg and Sharma (2022) and Cai et al. (2022) note that
the absence of expert involvement may compromise the reliability of recommendations.
Secondly, there is an administrative control gap, where existing systems lack mechanisms for
updating rules and guidelines as new nutritional knowledge emerges. This limits the system’s
ability to remain current and accurate.
Thirdly, there is a challenge in balancing personalization and medical safety. While some
systems prioritize user preferences, they may overlook important clinical constraints, leading to
potentially unsafe recommendations.
To address these gaps, this study proposes a dual-interface expert system that incorporates both
user interaction and professional oversight. By integrating forward chaining reasoning with an
administrative dashboard, the system ensures that recommendations are personalized,
transparent, and continuously updated based on expert input.

REFERENCES
Abaidi, A., Moussaoui, A., & Kasirzadeh, A. (2022). Constraint-based recommendation systems
for specialized medical diets. Journal of AI in Healthcare, 14(2), 102–118.
Ahmed, E., Oumer, M., & Hassan, M. (2025). Diabetes-focused food recommender system
(DFRS) to enabling digital health. PLOS Digital Health, 4(2), e0000721.
Butti, G., & Gangappa, M. (2020). Nutrition diet recommendation system using user's interest.
International Journal of Advanced Research in Engineering and Technology, 11(12), 2910–2919.
Cai, Y., Yu, F., Kumar, M., Gladney, R., & Mostafa, J. (2022). Health recommender systems
development, usage, and evaluation from 2010 to 2022: A scoping review. International Journal
of Environmental Research and Public Health, 19(22), 15115.
Durkin, J. (2024). Expert systems: Design and development. Macmillan Publishing Company.
Fandale, S., Kulkarni, P., & Patil, S. (2023). Diet recommendation system based on physical
activity levels and BMR. International Journal of Computer Applications, 185(10), 34–40.
Fisichella, M., Stewart, A., & Henze, N. (2025). Nutri-Graph: Leveraging graph neural networks
and Nutri-Score for healthy food recommendation. Applied Sciences, 15, 9283.
French, C. S. (2022). CIT474: Introduction to expert systems. National Open University of
Nigeria.
Garg, R., & Sharma, S. (2022). Analysis of user-centric vs. professional-centric dietary tools.
Journal of Software Engineering and Applications, 15(4), 88–102.
Hadi, A. S. (2024). Basic concepts of expert systems: Knowledge representation and inference.
Al-Mustansiriya University Publication.
Ikawati, Y., & Ramadhan, B. A. (2026). Interactive web-based expert system for personalized
diet and exercise recommendations using forward chaining. Journal of Advanced Vocational
Information and Communication Technology, 1(1), 1–14.
Kalpakoglou, K., Calderón-Pérez, L., Boqué, N., Guldas, M., Erdoğan Demir, Ç., Gymnopoulos,
L. P., & Dimitropoulos, K. (2025). An AI-based nutrition recommendation system technical
validation with insights from Mediterranean cuisine. Frontiers in Nutrition, 12, 1546107.
Kumari, D. N. N., Satya, T. P., Manikanta, B., Chandana, A. P., & Aditya, Y. L. (2024).
Personalized diet recommendation system using machine learning. Godavari Institute of
Engineering and Technology Research.
Lambay, M. A., & Mohideen, S. P. (2022). A hybrid approach-based diet recommendation
system using ML and big data analytics. Journal of Mobile Multimedia, 18(6), 1535–1562.
Lucas, P. J. F., & van der Gaag, L. C. (1991). Principles of expert systems. Addison-Wesley.
Papastratis, I., Konstantinidis, D., Daras, P., & Dimitropoulos, K. (2024). AI nutrition
recommendation using a deep generative model and ChatGPT. Scientific Reports, 14, 14620.
Stefanidis, K., Tsatsou, D., Konstantinidis, D., Gymnopoulos, L., Daras, P., Wilson-Barnes, S., &
Dimitropoulos, K. (2022). PROTEIN AI Advisor: A knowledge-based recommendation
framework using expert-validated meals for healthy diets. Nutrients, 14(20), 4435.
Tsolakidis, D., Gymnopoulos, L. P., & Dimitropoulos, K. (2024). Artificial intelligence and
machine learning technologies for personalized nutrition: A review. Informatics, 11(3), 62.
Vandeputte, J., Herold, P., Kuslii, M., Viappiani, P., Muller, L., Martin, C., & Cornuéjols, A.
(2023). Principles and validations of an artificial intelligence-based recommender system
suggesting acceptable food changes. The Journal of Nutrition, 153, 598–604.
Wang, X., Sun, Z., Xue, H., & An, R. (2025). Artificial intelligence applications to personalized
dietary recommendations: A systematic review. Healthcare, 13(12), 1417.
Yera, R., Alzahrani, A. A., Martínez, L., & Rodríguez, R. M. (2023). A systematic review on
food recommender systems for diabetic patients. International Journal of Environmental
Research and Public Health, 20(5), 4248.