Article: Knowledge graph construction and GCN prediction model for tort liability elements in the Civil Code Journal: International Journal of Reasoning-based Intelligent Systems (IJRIS) 2026 Vol.18 No.7 pp.1 - 11 Abstract: This paper addresses the need for intelligent analysis of tort liability elements under the Civil Code by proposing a legal judgement prediction method that integrates knowledge graphs with graph neural networks. By constructing a knowledge graph of tort liability elements, this study proposes an element alignment method combining Laplace coding with attention mechanisms to precisely link factual circumstances with legal elements. Building upon this foundation, an end-to-end multi-task graph convolutional network prediction model was designed to simultaneously perform liability determination and identification of specific element statuses. Experiments on public datasets such as CAIL2018-Small demonstrate that this method achieves an accuracy of 89.7% and a Macro-F1 score of 88.5%, significantly enhancing both predictive performance and interpretability. This research provides a reliable technical pathway for intelligent judicial assistance systems and holds positive implications for advancing judicial intelligence. Inderscience Publishers - linking academia, business and industry through research

Title: Knowledge graph construction and GCN prediction model for tort liability elements in the Civil Code

Authors: Binjing Li

Addresses: Department of Law and Literature, Harbin Finance University, Heilongjiang, 150030, China

Abstract: This paper addresses the need for intelligent analysis of tort liability elements under the Civil Code by proposing a legal judgement prediction method that integrates knowledge graphs with graph neural networks. By constructing a knowledge graph of tort liability elements, this study proposes an element alignment method combining Laplace coding with attention mechanisms to precisely link factual circumstances with legal elements. Building upon this foundation, an end-to-end multi-task graph convolutional network prediction model was designed to simultaneously perform liability determination and identification of specific element statuses. Experiments on public datasets such as CAIL2018-Small demonstrate that this method achieves an accuracy of 89.7% and a Macro-F1 score of 88.5%, significantly enhancing both predictive performance and interpretability. This research provides a reliable technical pathway for intelligent judicial assistance systems and holds positive implications for advancing judicial intelligence.

Keywords: knowledge graph; GCN; tort liability elements; the Civil Code.

DOI: 10.1504/IJRIS.2026.151416

International Journal of Reasoning-based Intelligent Systems, 2026 Vol.18 No.7, pp.1 - 11

Received: 13 Oct 2025
Accepted: 05 Nov 2025
Published online: 28 Jan 2026 *

Title: Knowledge graph construction and GCN prediction model for tort liability elements in the Civil Code

Keep up-to-date