<p>Artificial intelligence systems have demonstrated considerable proficiency in explicating concepts through textual means; however, their capacity for effective visual knowledge representation remains limited. This paper proposes a novel system that integrates Retrieval-Augmented Generation (RAG) architecture with advanced entity and relationship extraction methodologies to enable the visualization of computer science concepts. The RAG-based architecture employed comprises of three integral components: a reader module responsible for encoding textual content, a retriever module designed to identify and fetch contextually relevant document passages, and a generator module that synthesizes responses grounded in the retrieved information. Empirical evaluation indicates that this RAG-enabled framework outperforms direct large language model (LLM) querying in terms of response accuracy. This work presents a comparative analysis of two entity-relationship (ER) extraction paradigms: a rule-based (RB) approach leveraging syntactic and semantic pattern matching, and an LLM-based approach utilizing attention mechanisms inherent to transformer architectures to extract salient entities and their interrelations. The resultant ER graphs facilitate exploratory analysis of conceptual linkages within the domain. The core contributions reside in a systematic investigation of how critical parameters—including LLM model selection, RAG chunk size, and ER extraction strategies—influence the efficacy of the knowledge visualization pipeline. The system produces concise, informative visual graphs that distill complex information into essential elements, thereby advancing pedagogical techniques in concept summarization and knowledge representation.</p>