Knowledge Graph-Powered Decentralized Personalization
  • Home
  • Schedule
  • Materials
  1. Schedule

On this page

  • Tutorial Schedule
    • Detailed Schedule
  • Prerequisites
  • Materials Needed

Schedule

Tutorial Schedule

Detailed Schedule

Duration Activity
10 min Introduction
Motivation, limitations of centralized personalization, and overview of PKG-driven decentralization.
30 min Personal Knowledge Graph Foundations
PKG representation, provenance, schema alignment, and examples (FoodKG, Health and Finance PKGs).
30 min Decentralized AI Architectures
Federated learning variants; blockchain-based smart contract coordination.
10 min Break
20 min Semantic–Neural Personalization
Rule-guided KG adaptation, behavioral alignment, and filter-bubble mitigation.
20 min Knowledge-Graph-Grounded RAG Pipelines
Parallel and multi-stage retrieval; building compact personalized contexts.
60 min Hands-On Exercise
Participants use provided Jupyter notebooks to:
• Build a small PKG using RDFLib
• Run a toy federated personalization flow using Flower
• Execute KG-based retrieval for LLM prompting
30 min Discussion and Q&A

Total Duration: 3 hours 20 minutes

NoteNote

The schedule includes interactive hands-on exercises. Participants are encouraged to bring a laptop with the required software pre-installed (see Materials Needed section below).


Prerequisites

  • Basic understanding of knowledge graphs
  • Familiarity with recommendation systems (helpful but not required)
  • Programming experience with Python (for hands-on session)

Materials Needed

  • Laptop with internet connection
  • Pre-installed software:
    • Python 3.8 or higher
    • Jupyter Notebook or JupyterLab
    • Git (for cloning repositories)
  • Required Python packages:
    • RDFLib - for PKG construction
    • Flower - for federated learning
    • Additional packages will be listed in the tutorial materials