Legal judgment prediction via dynamic task graph and keyword-aware contrastive learning
Jiawei Li, Shengjia Zhang · 2026
Read Legal judgment prediction via dynamic task graph and keyword-aware contrastive learning by Jiawei Li, Shengjia Zhang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0357463.txt.
Description
Jiawei Li and Shengjia Zhang published a 2026 paper in PLOS ONE on legal judgment prediction, teaching computers to forecast court outcomes from the written facts of criminal cases. A full prediction has three linked parts: which law articles apply, what the charge is, and how long the penalty runs. Earlier systems ignored the telling keywords in case descriptions, or treated the three parts as independent when they guide each other.
Their model, DTG-KCL, builds a small relationship graph for each case so the subtasks can inform one another case by case. A keyword mechanism, sharpened by contrastive learning, helps the model separate charges that look confusingly alike. On two real legal datasets, the system beat five comparison methods across the subtasks. Their code is released openly, as decision support research showing how structure inside legal tasks can be exploited rather than flattened.
Catalog record
| CREATOR | Jiawei Li, Shengjia Zhang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51086 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |