Legal Judgment Prediction (LJP) predicts applicable law articles, charges, and penalty terms from case facts. Beyond accuracy, LJP calls for intrinsically interpretable and legally grounded reasoning that can reconcile statutory rules with precedent-informed standards. However, existing methods often behave as static, one-shot predictors, providing limited procedural support for verifiable reasoning and little capability to adapt as jurisprudential practice evolves. We propose VERDICT, a self-refining collaborative multi-agent framework that simulates a virtual collegial panel. VERDICT assigns specialized agents to complementary roles (e.g., fact structuring, legal retrieval, opinion drafting, and supervisory verification) and coordinates them in a traceable draft--verify--revise workflow with explicit Pass/Reject feedback, producing verifiable reasoning traces and revision rationales. To capture evolving case experience, we further introduce a Hybrid Jurisprudential Memory (HJM) grounded in the Micro-Directive Paradigm, which stores precedent standards and continually distills validated multi-agent verification trajectories into updated Micro-Directives for continual learning across cases. We evaluate VERDICT on CAIL2018 and a newly constructed CJO2025 dataset with a strict future time-split for temporal generalization. VERDICT achieves state-of-the-art performance on CAIL2018 and demonstrates strong generalization on CJO2025. To facilitate reproducibility and further research, we release our code and the dataset at https://anonymous.4open.science/r/ARR-4437.
Figures & tables
Figure 1: The overall inference framework of VERDICT. It illustrates the interaction between the Traceable Multi-Agent Workflow and the Hybrid Jurisprudential Memory (HJM) . The process evolves through Preparation, Drafting & Review, and Final Adjudication phases. Note that the Case Judge agent is instantiated using our domain-specific aligned expert model.
Figure 2: The evolutionary mechanism of the Hybrid Jurisprudential Memory. The lifecycle strategy Φtrans transforms archived empirical Standards into precise and compact Micro-Directives through three phases.
Law Article
Charge
Term of Penalty
Acc%
MP%
MR%
MF1%
Hit@2
Acc%
MP%
MR%
MF1%
Hit@2
Acc%
MP%
MR%
MF1%
1
CNN
69.52
62.14
58.35
60.18
74.15
73.56
70.45
65.23
67.74
79.66
33.24
28.56
25.14
26.74
2
BERT
76.54
73.21
70.45
71.80
81.87
75.85
78.45
76.12
77.26
82.25
29.80
32.14
29.56
30.79
3
MLAC
73.02
69.27
66.14
64.23
79.54
74.73
72.65
69.56
68.36
81.27
36.45
34.50
29.95
29.64
4
TopJudge
78.60
76.59
74.84
73.72
84.31
81.17
81.87
80.57
79.96
86.13
35.70
32.81
31.03
31.49
5
MPBFN
76.83
74.57
71.45
70.57
82.29
76.17
78.88
75.65
75.68
84.33
36.18
33.67
30.08
29.43
Table 1: Performance Comparisons on CAIL2018. The best results are bolded and the second-best are underlined
Law Article
Charge
Term of Penalty
Acc%
MP%
MR%
MF1%
Hit@2
Acc%
MP%
MR%
MF1%
Hit@2
Acc%
MP%
MR%
MF1%
1
TopJudge
71.30
68.89
74.2
71.28
80.35
72.19
71.82
69.45
70.53
83.26
36.83
27.45
24.96
26.10
2
EPM
75.23
76.45
74.89
76.10
87.77
73.42
74.15
69.88
72.20
85.72
43.44
25.10
28.77
26.65
3
LADAN
71.23
72.45
70.98
71.12
85.70
70.23
71.45
69.89
70.67
84.10
37.92
28.32
27.70
27.14
4
NeurJudge
73.69
74.40
80.14
72.44
84.20
74.30
71.59
77.5
70.56
86.17
42.87
25.40
26.37
25.14
5
CTM
76.71
68.30
71.07
69.57
83.18
73.17
72.25
71.40
70.31
85.18
41.17
24.50
26.67
25.33
Table 2: Robustness evaluation on CJO2025. The best results are bolded and the second-best are underlined
Method
CJO2025
CAIL2018
Law Article
Charge
Prison Term
Law Article
Charge
Prison Term
Acc
Ma-F
Acc
Ma-F
Acc
Ma-F
Acc
Ma-F
Acc
Ma-F
Acc
Ma-F
Vanilla
82.12
76.54
77.61
72.10
27.06
19.76
73.04
72.01
67.95
69.13
26.07
25.01
Vanilla+
83.08
78.83
79.21
71.68
28.17
22.18
76.40
75.52
70.80
72.18
24.14
25.25
w/o Memory
86.55
83.70
81.30
79.40
44.71
24.10
82.60
82.10
79.13
81.50
35.77
30.20
w/o MAS
87.32
85.42
83.48
80.32
39.50
30.15
83.10
81.50
79.80
82.30
36.20
28.10
Table 3: Ablation experiments. The best results are bolded and the second-best are underlined .
Input Case (Sample from CJO2025)
On January 15, 2025, defendant Zhang was drinking at a public BBQ stall . Due to a trivial conflict (stranger Li looked at him), Zhang felt provoked. Zhang beat Li with a beer bottle , resulting in chaos at the stall. Forensic examination confirmed that Li suffered a Minor Injury . Zhang claimed he was just venting anger and had no personal grudge against Li.
Vanilla
PLJP
VERDICT (Ours)
Logic: Keyword Matching The model focuses on high-frequency patterns without deep reasoning. Internal Monologue: “Detected keywords ‘beat’ and ‘Minor Injury’ . In training data, ‘Minor Injury’ strongly correlates with Art. 234.”
Logic: Precedents Retrieval Retrieves precedents but lacks systematic adjudication experience. Reasoning Process: “Retrieved precedents labeled Art. 234 and Art. 293. However, since the input’s objective result (‘Minor Injury’) shows high textual similarity with Art. 234 cases, the model aligns with them and ignores the motive nuance in Art. 293 precedents.”
Logic: Retrieval + Refinement Uses Micro-Directives & Supervisor . Agent Interaction Loop: 1. Assistant: Retrieves directive “Public Place + Venting Anger → Art. 293”. 2. Case Judge: Drafts Art. 234 based on injury result. 3. Supervisor: REJECTS draft. “Logic Error: Location is public ; Motive is provocation . Social order violation > Personal injury. Apply Art. 293.”
Wrong Prediction ✗ Art. 234 (Intentional Injury)
Wrong Prediction ✗ Art. 234 (Intentional Injury)
Correct Judgment ✓ Art. 293 (Picking Quarrels)
Figure 3: Case study comparison. The red highlights indicate misleading surface features (resulting in Intentional Injury), while the green highlights denote contextual evidence supporting Picking Quarrels. Vanilla falls into the keyword trap; PLJP fails to resolve the statutory conflict; VERDICT correctly identifies the crime’s nature via the Supervisor’s logical rectification.
Appendix figures & tables6 assets
Supplementary material from the paper’s appendix.
Appendix
Type
CAIL2018
CJO2025
# Law Article
99
67
# Charge
115
73
# Prison Term
11
11
# Sample
134739
8199
Avg. # words in Fact
288.6
332.13
Appendix
Table 4: Statistics of the datasets.
Figure 4: System prompt for the Court Clerk Agent , responsible for distilling objective event points from raw facts.
Figure 5: System prompt for the Judicial Assistant Agent , performing semantic re-ranking of precedents and statutes.
Figure 6: System prompt for the Case-handling Judge Agent . It includes logic for initial drafting and iterative refinement based on feedback.
Figure 7: System prompt for the Adjudication Supervisor Agent , responsible for logical consistency checks and issuing correction signals.
Figure 8: System prompt for the Presiding Judge Agent , synthesizing the multi-agent workflow into a standard verdict.