Exploring Semantic Labeling Strategies for Third-Party Cybersecurity Risk Assessment Questionnaires

Researchers developed semantic labeling strategies to improve third-party cybersecurity risk assessment questionnaires using Large Language Models and semi-supervised learning. The study demonstrates that semantic labels can enhance question retrieval for cybersecurity assessments while reducing LLM costs through hybrid approaches.

• →Traditional TPRA questionnaire selection relies on manual processes and keyword matching, which often misses semantic meaning.
• →Semantic labeling using LLMs can improve alignment between cybersecurity assessment needs and question retrieval.
• →Semi-supervised semantic labeling (SSSL) reduces LLM usage costs while maintaining labeling quality across large question repositories.
• →The hybrid approach clusters questions in embedding space and propagates labels using k-Nearest Neighbors for efficiency.
• →Discriminative and consistent semantic labels are key to improving downstream retrieval performance in cybersecurity assessments.