Legal knowledge graphs can represent legal entities‚ concepts‚ documents‚ provisions‚ cases and their relationships in a form that can be queried and interpreted․ In this paper‚ we provide a systematic overview of legal knowledge graphs in legal decision-making processes‚ focusing on their construction‚ representation‚ and application in various downstream tasks․ Accordingly‚ we study the following tasks: predicting legal decisions‚ predicting legal charges‚ legal case retrieval‚ legal statute identification‚ legal question answering‚ legal compliance checking‚ and explainable legal reasoning․ Besides‚ we summarize the pipelines of constructing legal knowledge graphs into three categories‚ namely‚ ontology-based manually constructed legal knowledge graphs‚ knowledge graphs constructed by the NLP techniques‚ and legal knowledge graph constructed with the help of LLMs․ The review identifies the potential of graph-based models to improve retrieval‚ prediction‚ matching‚ and explanation of legal concepts compared to flat-text and keyword-based models․ This article presents them as knowledge infrastructures to improve access‚ traceability‚ consistency‚ oversight‚ automation‚ and transparency of legal decision-support systems․
Keywords: Artificial Intelligence, explainable legal reasoning., knowledge graphs, large language models, legal decision-making
