With the rapid advancement of digital transformation and artificial intelligence technologies, the importance of data security has become increasingly prominent. Among these, the compliance evaluation of cryptographic schemes has emerged as a critical defense line for safeguarding data security due to its significance. The core challenge lies in translating ambiguous standards described in natural language into logical assertions that can be processed by computers. Among existing approaches, manual evaluation suffers from inefficiency and incomplete coverage; rule-based methods lack deep modeling of domain concepts; and machine learning methods offer good scalability but struggle with interpretability. While knowledge graph technology holds potential to integrate the strengths of these approaches, its application in cryptographic scheme compliance evaluation remains an under-explored domain. Addressing these issues, this paper proposes KG-Engine, an automated evaluation framework based on ontology-driven knowledge graphs. This framework leverages a formalized ontology of cryptographic misuse and compliance evaluation to construct a precise semantic framework. Within this framework, declarative mapping rules transform raw data into structured knowledge graphs. Based on this graph, automated and explainable compliance detection is achieved through logical reasoning mechanisms involving entity linking, graph traversal, context awareness, and risk attribution. Experimental results demonstrate that KG-Engine significantly outperforms traditional baseline methods in detection precision, recall, and F1 score, while effectively identifying compound risks. This provides an effective solution for advancing data security compliance evaluation from “human-driven” to “intelligence-driven.”