Daily Arxiv

This is a page that curates AI-related papers published worldwide.
All content here is summarized using Google Gemini and operated on a non-profit basis.
Copyright for each paper belongs to the authors and their institutions; please make sure to credit the source when sharing.

Enhancing Natural Language Inference Performance with Knowledge Graph for COVID-19 Automated Fact-Checking in Indonesian Language

Created by
  • Haebom

Author

Arief Purnama Muharram, Ayu Purwarianti

Outline

This paper proposes a novel model utilizing knowledge graphs (KGs) for automated COVID-19 information verification in Indonesian. To overcome the performance limitations of existing deep learning-based Natural Language Inference (NLI) methods, we focus on improving NLI performance by leveraging KGs as external knowledge. The proposed model consists of three modules: a fact module, an NLI module, and a classifier module. It processes information from the KGs and processes the semantic relationships between given premises and hypotheses to derive the final result. Training using the Indonesian COVID-19 information verification dataset and the COVID-19 KG Bahasa Indonesia, we achieved an accuracy of 0.8616, demonstrating the effectiveness of utilizing KGs.

Takeaways, Limitations

•
Takeaways:
◦
We demonstrate that knowledge graphs (KGs) can be used to improve the performance of natural language inference (NLI)-based automated information verification systems.
◦
Suggesting the possibility of building an effective information verification system using KG even in low-resource language environments such as Indonesian.
◦
It has high applicability not only to COVID-19 information verification but also to the development of automated information verification systems in other fields.
•
Limitations:
◦
Performance may be affected by the size and quality of the dataset used.
◦
Additional consideration may be needed for linguistic features specific to Indonesian.
◦
The completeness and quality of the KG directly impacts model performance. Inaccurate or incomplete information in the KG can lead to errors.
◦
Further research is needed to determine generalizability to other languages.
👍
Made with Slashpage