GpxCP: Uncertainty-Aware Explainable Recommendation via Graphex Calibration and Conformal Prediction

---

Overview

GpxCP proposes an uncertainty-aware and explainable recommendation framework built upon Graphex Calibration and Conformal Prediction. This repository contains the official implementation and experiment scripts used in the study.

---

Requirements & Setup

This project is developed using Python 3.10+. A virtual environment is recommended to isolate dependencies.

1️⃣ Clone the Repository

Clone the repository from our anonymous review link:

git clone https://anonymous.4open.science/r/GpxCP-3222
cd GpxCP

---

2️⃣ Create and Activate a Virtual Environment

python3 -m venv venv
source venv/bin/activate      # for Linux / macOS
venv\Scripts\activate         # for Windows

---

3️⃣ Install Dependencies

pip install --upgrade pip
pip install -r requirements.txt

This installs all necessary dependencies for running GpxCP.

---

Datasets

We utilize four widely used real-world datasets for evaluation: MovieLens, Amazon-Book, DBLP, and Wiki.
All datasets are preprocessed and included in the data/ directory of this repository.

Dataset	Description	Path
MovieLens	User–movie interaction dataset	data/movielens/
Amazon-Book	User–item interactions from Amazon	data/amazon-book/
DBLP	Co-authorship and citation network	data/dblp/
Wiki	User–page editing dataset	data/wiki/

---

Test

Testing Recommendation in Section 6.2.1

python test_rs.py

Testing Uncertainty Quantification in Section 6.2.2

python test_uq.py

---

Project Structure

GpxCP/
│
├── data/                # Datasets (MovieLens, Amazon-Book, DBLP, Wiki)
├── utils/               # Utility functions
├── figures/             # Methodology and result figures
├── test_rs.py           # Main test: recommendation
├── test_uq.py           # Main test: uncertainty quantification
├── requirements.txt     # Python dependencies
└── README.md

---

Thanks to

We thank the authors of the following open-source libraries for their foundational contributions:

• LightGCN
• Conformal Prediction

Their work provided essential resources for our implementation and experiments.

---

Citation

If you find this work useful, please cite:

---