Integrating collaborative filtering and chatbot optimization for personalized cybersecurity training

Details

Supervisors

Sadre, Ramin ; Duhoux, Benoît

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] en sciences informatiques, à finalité spécialisée

Abstract

enCybersecurity training increasingly relies on cyber ranges, interactive simulation platforms designed to provide hands-on and realistic learning experiences. However, these environments often lack personalized and pedagogically aligned guidance, which can lead to reduced learner engagement and suboptimal outcomes. This thesis addresses this issue by enhancing an existing gamified cybersecurity training platform with two upgraded, AI-driven components: a hybrid recommender system and a pedagogically informed chatbot coach. The proposed recommender system integrates collaborative filtering to form a hybrid recommender system to generate personalized scenario suggestions tailored to each learner’s profile. To enable meaningful evaluation, a synthetic dataset was constructed using a statistical simulation pipeline that models user behavior and calibrates scenario metadata. In parallel, the platform’s chatbot was enhanced using new prompt engineering techniques and new input/output filtering layers to improve pedagogical aspect of the coach. This chatbot provides spoiler-free, context-aware guidance aimed at promoting autonomous problem solving while avoiding direct solution disclosure. An empirical study involving nine participants demonstrates that the system improves the educational value of chatbot interactions. The results support the potential of AI-enhanced systems to deliver more effective and engaging cybersecurity training experiences.