PHaSE Project

Promoting Healthy and Sustainable Eating through Interactive and Explainable AI Methods

PRIN 2022 – Prot. 2022H5X7BK | CUP: H53D23003530006 | 24 Months
Funded by the Italian Ministry of University and Research (MUR)

Project Overview

PHaSE aimed to advance the state of the art in AI-based food recommender systems by integrating conversational agents, explainable AI, and persuasive dialogue strategies to foster healthier and more sustainable eating behaviors.

The project combined recommender systems, question answering over knowledge graphs, user modeling, and natural language generation into a fully working conversational agent. The system supports users in identifying healthy and sustainable recipes, understanding nutritional and environmental impact, and discovering substitute ingredients and meals.

Scientific Objectives

Food Recommender System

Development of a knowledge-based food recommender system suggesting healthy and sustainable recipes tailored to users’ preferences, physiological characteristics and goals.

Question Answering

Implementation of a QA component to answer natural language questions about ingredients, recipes, sustainability and identification of healthier substitutes.

Persuasive Dialogue

Design of persuasive natural language strategies grounded in behavioral theories to enhance awareness, transparency, and trust in AI-driven recommendations.

Conversational Agent

Integration of RS, QA, NLU and NLG modules into a fully working conversational agent evaluated “in the wild” with users.

Consortium

UNIBA

Coordinating Unit

Principal Investigator: Prof. Cataldo Musto

Expertise: Natural Language Processing, Conversational Agents, Explainable AI.

Role: Responsible for Dialogue, NLU, NLG and system integration.

UNICA

Partner Unit

Unit Leader: Prof. Ludovico Boratto

Expertise: Recommender Systems, Knowledge Graphs, Sustainability-aware algorithms.

Role: Responsible for Food RS, data modeling and dissemination.

UNITO

Partner Unit

Unit Leader: Prof. Amon Rapp

Expertise: Human–Computer Interaction, Co-design, User Evaluation.

Role: Responsible for participatory design and evaluation studies.

Dissemination

Scientific Publications

⯈ E-Mealio: An LLM-Powered Conversational Agent for Sustainable and Healthy Food Recommendation
Iacovazzi, A.R., Blanco, L., Spillo, G., Musto, C. (2026).
⯈ Instructing and Prompting Large Language Models for Explainable Cross-domain Recommendations
A. Petruzzelli, C. Musto, L. Laraspata, I. Rinaldi, M. de Gemmis, P. Lops, G. Semeraro
⯈ Recommending Healthy and Sustainable Meals exploiting Food Retrieval and Large Language Models
A. Petruzzelli, C. Musto, M.C. Di Carlo, G. Tempesta, G. Semeraro
⯈ HeASe: An AI-powered Framework to Promote Healthy and Sustainable Eating
A. Petruzzelli, C. Musto, M.C. Di Carlo, G. Tempesta, G. Semeraro
⯈ FoodNexus: Massive Food Knowledge for Recommender Systems
Boratto, L., Fenu, G., Marras, M., Medda, G., & Zedda, G. (Published in ECIR 2026)
⯈ hopwise: A Python Library for Explainable Recommendation based on Path Reasoning over Knowledge Graphs
Boratto, L., Fenu, G., Marras, M., Medda, G., & Soccol, A.
⯈ GreenFoodLens: Sustainability Labels for Food Recommendation
Balloccu, G., Boratto, L., Fenu, G., Marras, M., & Murgia, G.
⯈ Fair augmentation for graph collaborative filtering
Boratto, L., Fabbri, F., Fenu, G., Marras, M., & Medda, G.
⯈ Comprehensive Assessment of Robustness in Fairness of GNN-based Recommender Systems against Attacks
Boratto, L., Fabbri, F., Fenu, G., Marras, M., & Medda, G.
⯈ PHaSE Project - Promoting Healthy and Sustainable Eating through Interactive and Explainable AI Methods
Musto, C., Rapp, A., and Boratto, L. (2025).
⯈ Designing for Healthy Food Practices: Challenges and Opportunities for Changing People’s Eating Behavior Using Persuasive Technology
Rapp, A., Boldi, A. (2025).
⯈ Promoting Healthy Eating by Design: Opportunities for Meaningful Persuasive Technologies
Rapp, A., Boldi, A. (2025).

Open Science & Impact

PHaSE adopted an Open Science approach, releasing datasets, knowledge graphs, APIs and software components. The project contributes to sustainable food consumption, ecological transition and AI transparency.

Comprehensive Assessment of Robustness in Fairness of GNN-based Recommender Systems against Attacks

Source Code

Events

Project dissemination included presentations at several national and international conferences:

✓ RecSys 2024 & 2025 (ACM Conference on Recommender Systems)
✓ ECIR 2025 & 2026 (European Conference on Information Retrieval)
✓ IJCAI 2024 (Int. Joint Conference on Artificial Intelligence - Workshop)
✓ RecSoGood 2025