The interpretability of AI is just as important as its performance. In the LegalAI field, there have been efforts to enhance the interpretability of models, but a trade-off between interpretability and prediction accuracy remains inevitable. In this paper, we introduce a novel framework called LK-IB for compulsory measure prediction (CMP), one of the critical tasks in LegalAI. LK-IB leverages Legal Knowledge and combines an Interpretable model and a Black-box model to balance interpretability and prediction performance. Specifically, LK-IB involves three steps: (...) (1) inputting cases into the first module, where first-order logic (FOL) rules are used to make predictions and output them directly if possible; (2) sending cases to the second module if FOL rules are not applicable, where a case distributor categorizes them as either “simple” or “complex“; and (3) sending simple cases to an interpretable model with strong interpretability and complex cases to a black-box model with outstanding performance. Experimental results demonstrate that the LK-IB framework provides more interpretable and accurate predictions than other state-of-the-art models. Given that the majority of cases in LegalAI are simple, the idea of model combination has significant potential for practical applications. (shrink)

Previous studies on crisis know­ledge organization mostly focused on the categorization of crisis know­ledge without regarding its dynamic trend and temporal-spatial features. In order to emphasize the dynamic factors of crisis collaboration, a fine-grained crisis know­ledge model is proposed by integrating temporal-spatial analysis based on ontology, which is one of the commonly used methods for know­ledge organization. The reconstruction of ontology-based crisis know­ledge will be implemented through three steps: analyzing temporal-spatial features of crisis know­ledge, reconstructing crisis know­ledge ontology, and verifying (...) the temporal-spatial ontology. In the process of ontology reconstruction, the main classes and properties of the domain will be identified by investigating the crisis information resources. Meanwhile the fine-grained crisis ontology will be achieved at the level of characteristic representation of crisis know­ledge including temporal relationship, spatial relationship, and semantic relationship. Finally, we conducted case addition and system implementation to verify our crisis know­ledge model. This ontology-based know­ledge organization method theoretically optimizes the static organizational structure of crisis know­ledge, improving the flexibility of know­ledge organization and efficiency of emergency response. In practice, the proposed fine-grained ontology is supposed to be more in line with the real situation of emergency collaboration and management. Moreover, it will also provide the know­ledge base for decision-making during rescue process. (shrink)