Re-Identification in Court Decisions with Knowledge Graphs


We propose to approach the general problem of re-identification (not only for a specific domain) with knowledge graphs (ontologies). In a first step, you will construct knowledge graphs from both external data and court decisions. Then you will perform ontology alignment [2] to connect the two knowledge graphs (a big one from external data and a small one from the court decision to be re-identified). An anonymized node from the knowledge graph of the court decision being aligned to a node from the external data knowledge graph is a re-identification. In a last step, you will analyze the success of your system by measuring the recall and precision of the re-identifications.