EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Towards Interpretable Embeddings: Aligning Representations With Semantic Aspects

Nitisha Jain, Antoine Domingues, Adwait Baokar, Albert Meroño Peñuela, Elena Simperl

Neurosymbolic Artificial Intelligence · 2025

DOI: 10.1177/29498732251377351 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Neurosymbolic Artificial Intelligence, 2025. Verfasst von Nitisha Jain und 4 weiteren Autor:innen. Die Arbeit nennt 64 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Knowledge graph embedding models (KGEMs) project entities and relations from knowledge graphs (KGs) into dense vector spaces, enabling tasks such as link prediction and recommendation systems. However, these embeddings typically suffer from a lack of interpretability and struggle to represent entity similarities in a way that is meaningful to humans. To address these challenges, we introduce InterpretE , a neuro-symbolic approach that generates interpretable vector spaces aligned with human-understandable entity aspects. By explicitly linking entity representations to their desired semantic aspects, InterpretE not only improves interpretability but also enhances the clustering of similar entities based on these aspects. Our experiments demonstrate that InterpretE effectively produces embeddings that are interpretable and improve the evaluation of semantic similarities, making it a valuable tool in explainable AI research by supporting transparent decision-making. By offering insights into how embeddings represent entities, InterpretE enables KGEMs to be used for semantic tasks in a more trustworthy and reliable manner.

Zitieren

APA

Jain, N., Domingues, A., Baokar, A., Peñuela, A. M., & Simperl, E. (2025). Towards Interpretable Embeddings: Aligning Representations With Semantic Aspects. Neurosymbolic Artificial Intelligence. https://doi.org/10.1177/29498732251377351

Vancouver

Jain N, Domingues A, Baokar A, Peñuela AM, Simperl E. Towards Interpretable Embeddings: Aligning Representations With Semantic Aspects. Neurosymbolic Artificial Intelligence. 2025. doi:10.1177/29498732251377351

BibTeX
@article{Jain2025,
  author = {Jain, Nitisha and Domingues, Antoine and Baokar, Adwait and Peñuela, Albert Meroño and Simperl, Elena},
  title = {Towards Interpretable Embeddings: Aligning Representations With Semantic Aspects},
  journal = {Neurosymbolic Artificial Intelligence},
  year = {2025},
  doi = {10.1177/29498732251377351}
}
Kostenlos starten →