Legal judgment prediction via dynamic task graph and keyword-aware contrastive learning
- 1Department of Network Engineering, Henan University, School of Software, Kaifeng, China
- 2Department of Computer Science and Technology, Henan University, International School of Technology, Kaifeng, China
Abstract
Legal judgment prediction (LJP) aims to predict judgment results based on the factual descriptions of criminal cases and is gradually becoming a popular research topic in the legal field. Case factual descriptions contain a large amount of keyword information, and there are complex dependencies among the subtasks. For example, law article prediction can guide charge prediction and term of penalty prediction. However, most of the previous methods often failed to take advantage of the effective keywords that are commonly present in case descriptions, or failed to consider the interrelationships between sub-tasks. To address the shortcomings, we propose a multi-task legal judgment prediction framework based on Dynamic Task Graph and Keyword-aware Contrastive Learning, termed DTG-KCL. Specifically, we design a sample-level method of dynamic task relationship graph convolution to encode task relationships from the perspective of each input case description. Subsequently, we enhance the ability to distinguish confusing charges by developing a more profound and novel keyword fusion mechanism based on soft keyword representation and an approach of keyword-aware contrastive learning. Our model effectively improved the performance of the sub-tasks. Experimental results on two real legal datasets demonstrated that our model had a significant advantage over the five methods in terms of prediction performance. Code can be obtained from: https://github.com/YiLisiForever/DTG-KCL.
Related
- How effective are contrastive learning-based approaches for activity-cliff prediction?
- Heterogeneous Graph Contrastive Learning for Drug-Gene-Disease Motif Prediction
- Predicting miRNA–Disease Associations Based on Contrastive Learning of Linear and Nonlinear Features
- Revealing pan-cancer clonal niches of T cells from single-cell RNA sequencing using contrastive learning
- Pocket contrastive learning and conformation fusion for robust structure-conditioned 3D molecular generation under receptor conformational variation.
BibTeX
@article{li2026legal,
title = {{Legal judgment prediction via dynamic task graph and keyword-aware contrastive learning}},
author = {Jiawei Li and Shengjia Zhang},
journal = {PLOS One},
year = {2026},
month = sep,
doi = {10.1371/journal.pone.0357463},
url = {https://doi.org/10.1371/journal.pone.0357463},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - Legal judgment prediction via dynamic task graph and keyword-aware contrastive learning
AU - Jiawei Li
AU - Shengjia Zhang
PY - 2026
DA - 2026/09/10
JO - PLOS One
DO - 10.1371/journal.pone.0357463
UR - https://doi.org/10.1371/journal.pone.0357463
AB - Legal judgment prediction (LJP) aims to predict judgment results based on the factual descriptions of criminal cases and is gradually becoming a popular research topic in the legal field. Case factual descriptions contain a large amount of keyword information, and there are complex dependencies among the subtasks. For example, law article prediction can guide charge prediction and term of penalty prediction. However, most of the previous methods often failed to take advantage of the effective keywords that are commonly present in case descriptions, or failed to consider the interrelationships between sub-tasks. To address the shortcomings, we propose a multi-task legal judgment prediction framework based on Dynamic Task Graph and Keyword-aware Contrastive Learning, termed DTG-KCL. Specifically, we design a sample-level method of dynamic task relationship graph convolution to encode task relationships from the perspective of each input case description. Subsequently, we enhance the ability to distinguish confusing charges by developing a more profound and novel keyword fusion mechanism based on soft keyword representation and an approach of keyword-aware contrastive learning. Our model effectively improved the performance of the sub-tasks. Experimental results on two real legal datasets demonstrated that our model had a significant advantage over the five methods in terms of prediction performance. Code can be obtained from: https://github.com/YiLisiForever/DTG-KCL.
SN - 1932-6203
ER -