In this paper, we tackle the incremental maintenance of Datalog inference materialisation when the rule set can be updated. This is particularly relevant in the context of the Internet of Things and Edge computing where smart devices may need to reason over newly acquired knowledge represented as Datalog rules. Our solution is based on an adaptation of a stratification strategy applied to a dependency hypergraph whose nodes correspond to rule sets in a Datalog program. Our implementation supports recursive rules containing both negation and aggregation. We demonstrate the effectiveness of our system on real and synthetic data.

翻译：本文研究了在规则集可更新的情况下，如何增量维护Datalog推理物化结果。这一问题在物联网和边缘计算场景中尤为重要——智能设备需要基于新获取的、以Datalog规则形式表示的知识进行推理。我们的方案基于对依赖超图（其节点对应Datalog程序中的规则集）应用分层策略的改进实现，支持包含否定和聚合的递归规则。通过在真实与合成数据上的实验，我们验证了系统的有效性。