Towards an Approach for Project-Library Recommendation Based on Graph Normalization

Abhinav Jamwal, Sandeep Kumar

2025

Abstract

The performance of project-library recommendation systems depends on the choice of graph normalization techniques. This work explores two primary normalization schemes within a knowledge graph - enhanced project - library recommendation system: symmetric normalized Laplacian (SNL) and random walk normalized Laplacian (RWL). Experimental results show that RWL consistently delivers better performance in key metrics, including mean precision (MP), mean recall (MR), and mean F1 score (MF), particularly in sparse datasets. Although SNL performs well in denser datasets, its effectiveness decreases with increasing sparsity. Furthermore, loss curves for collaborative filtering (CF) and knowledge graph (KG) tasks indicate that RWL converges faster and shows greater stability. These findings establish RWL as a reliable technique for improving GNN-based recommendation systems, especially in sparse and complex project-library interaction scenarios.

Paper Citation

in Harvard Style

Jamwal A. and Kumar S. (2025). Towards an Approach for Project-Library Recommendation Based on Graph Normalization. In Proceedings of the 20th International Conference on Evaluation of Novel Approaches to Software Engineering - Volume 1: ENASE; ISBN 978-989-758-742-9, SciTePress, pages 120-127. DOI: 10.5220/0013351900003928

in Bibtex Style

@conference{enase25,
author={Abhinav Jamwal and Sandeep Kumar},
title={Towards an Approach for Project-Library Recommendation Based on Graph Normalization},
booktitle={Proceedings of the 20th International Conference on Evaluation of Novel Approaches to Software Engineering - Volume 1: ENASE},
year={2025},
pages={120-127},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013351900003928},
isbn={978-989-758-742-9},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 20th International Conference on Evaluation of Novel Approaches to Software Engineering - Volume 1: ENASE
TI - Towards an Approach for Project-Library Recommendation Based on Graph Normalization
SN - 978-989-758-742-9
AU - Jamwal A.
AU - Kumar S.
PY - 2025
SP - 120
EP - 127
DO - 10.5220/0013351900003928
PB - SciTePress