Legal Judgment Prediction via graph boosting with constraints

Suxin Tong, Jingling Yuan, Peiliang Zhang, Lin Li

Abstract: Highlights•Mitigated the multi-task multi-label challenge in legal judgment prediction.•A multi-perspective interactive encoder is developed for fact and label fusion.•A multi-graph attention consistency expert module is proposed for task consistency.•Extensive experiments show GJudge’s superiority in legal judgment prediction.