Semantic ethical glass box (SEGB): A knowledge graph–based framework for transparent robot behaviour analysis in social robotics

Scenarios involving Human-Robot Interaction are becoming increasingly prevalent across multiple domains. In such settings, reconstructing and explaining robots’ decisions and actions remain a complex and largely unsolved challenge. The Semantic Ethical Glass Box (SEGB) toolkit addresses this problem by providing an end-to-end software pipeline for the post-hoc analysis and auditing of robot behaviour. SEGB enables social robots to generate semantically structured logs that are integrated into a global Knowledge Graph (KG) system and later exploited for post-hoc analysis. To this end, SEGB provides software modules that support: (1) semantic log generation on robotic platforms, (2) safe log management and storage within the KG, and (3) automated report generation with enriched insights extracted from the KG.