A new article co-authored by Associate Professor Teeradaj Racharak has been published in the journal Knowledge-Based Systems. The paper, titled “Weight-aware Tasks for Evaluating Knowledge Graph Embeddings”, proposes a novel evaluation framework that introduces weight-aware tasks to more accurately assess the performance of knowledge graph embedding models.

While traditional benchmarks treat all data points equally, the proposed method takes into account the relative importance of facts and relationships in a knowledge graph. This allows for a more realistic and fine-grained evaluation of how well embedding models capture key semantics and structural properties, especially in large-scale or domain-specific knowledge graphs.

This work contributes to advancing the evaluation methodology for representation learning and reflects ongoing research at the Advanced Institute of So-Go-Chi (Convergence Knowledge) Informatics toward interpretable and reliable AI for knowledge systems.

Citation:
Kong Wei Kun, Xin Liu, Teeradaj Racharak, Guanqun Sun, Qiang Ma, Le-Minh Nguyen, Weight-aware Tasks for Evaluating Knowledge Graph Embeddings, Knowledge-Based Systems, 2025.
DOI: https://doi.org/10.1016/j.knosys.2025.113596