Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Responsible AI in Education: Hybrid Recommendation System for K-12 Students Case Study

Nazarii Drushchak, Vladyslava Tyshchenko, Nataliya Polyakovska

TL;DR

This paper presents a graph-based hybrid recommendation system for K-12 EdTech that combines static relationship graphs with dynamic student feedback and matrix factorization, aiming to deliver personalized learning opportunities while addressing bias. A fairness analysis framework assesses respondent reactions across protected groups, using metrics like $|P(g_j,t_i) - P(g_{j+1},t_i)|$ with a tolerance $ΔP$ and a minimum sample $N_{sample}$ to flag disparities, enabling ongoing transparency and monitoring. The case study in Mesquite ISD (AYO platform) demonstrates the approach with explicit transparency logs, reliability checks, and governance through Looker dashboards, while discussing limitations and future bias-mitigation strategies. The work contributes a practical, auditable workflow for responsible AI in education, highlighting the need for continuous fairness evaluation and stakeholder-inclusive monitoring to ensure equitable access to resources. Overall, the integration of graph-based modeling, matrix factorization, and fairness auditing provides a concrete blueprint for deploying equitable, transparent AI-driven education in real-world districts.

Abstract

The growth of Educational Technology (EdTech) has enabled highly personalized learning experiences through Artificial Intelligence (AI)-based recommendation systems tailored to each student needs. However, these systems can unintentionally introduce biases, potentially limiting fair access to learning resources. This study presents a recommendation system for K-12 students, combining graph-based modeling and matrix factorization to provide personalized suggestions for extracurricular activities, learning resources, and volunteering opportunities. To address fairness concerns, the system includes a framework to detect and reduce biases by analyzing feedback across protected student groups. This work highlights the need for continuous monitoring in educational recommendation systems to support equitable, transparent, and effective learning opportunities for all students.

Towards Responsible AI in Education: Hybrid Recommendation System for K-12 Students Case Study

TL;DR

This paper presents a graph-based hybrid recommendation system for K-12 EdTech that combines static relationship graphs with dynamic student feedback and matrix factorization, aiming to deliver personalized learning opportunities while addressing bias. A fairness analysis framework assesses respondent reactions across protected groups, using metrics like with a tolerance and a minimum sample to flag disparities, enabling ongoing transparency and monitoring. The case study in Mesquite ISD (AYO platform) demonstrates the approach with explicit transparency logs, reliability checks, and governance through Looker dashboards, while discussing limitations and future bias-mitigation strategies. The work contributes a practical, auditable workflow for responsible AI in education, highlighting the need for continuous fairness evaluation and stakeholder-inclusive monitoring to ensure equitable access to resources. Overall, the integration of graph-based modeling, matrix factorization, and fairness auditing provides a concrete blueprint for deploying equitable, transparent AI-driven education in real-world districts.

Abstract

The growth of Educational Technology (EdTech) has enabled highly personalized learning experiences through Artificial Intelligence (AI)-based recommendation systems tailored to each student needs. However, these systems can unintentionally introduce biases, potentially limiting fair access to learning resources. This study presents a recommendation system for K-12 students, combining graph-based modeling and matrix factorization to provide personalized suggestions for extracurricular activities, learning resources, and volunteering opportunities. To address fairness concerns, the system includes a framework to detect and reduce biases by analyzing feedback across protected student groups. This work highlights the need for continuous monitoring in educational recommendation systems to support equitable, transparent, and effective learning opportunities for all students.

Paper Structure

This paper contains 25 sections, 3 figures, 3 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Background and Related Work
  3. Personalization in Educational Technology
  4. Graph-Based Recommendation Systems
  5. Fairness in AI-Based Recommendation Systems
  6. Hybrid Approaches in Recommendation Systems
  7. Research Gaps and Problem Formulation
  8. Methodology
  9. Graph Recommendation System
  10. Fairness Analysis
  11. Case Study
  12. Solution Explanation
  13. Transparency
  14. Content-filtering part
  15. Collaborative-filtering part
  16. ...and 10 more sections

Figures (3)

  • Figure 1: Representation of Student Neighbourhood Subgraph Sample representation of induced subgraph of neighbors centered at Student node within a given radius. Student nodes (blue) represent each unique student and connect with other types of entities. Orange nodes represent a path for new suggestions (neighborhood of radius 3). A green node represents a new suggestion and new connection in the graph of type "Suggested". Each edge in the graph has type and weight (0;1] located under the edge label)
  • Figure 2: Element of the Recommendation System User Interface in the Solution Whenever a user reviews a new recommendation, the hover over the suggestion icon triggers an appearance of the tooltip with the reasoning behind the recommendation.
  • Figure 3: Recommendations Precision In Gender Groups The Protected group label (F/M) is displayed on the left side of the vertical axis. The target category is shown on the left side of the vertical axis. The horizontal axis of each chart provides recommendations for precision percentage.