Knowledge Graphs (KGs) serve as an effective approach for depicting organized knowledge, allowing for efficient responses to inquiries. Large Language Models have demonstrated remarkable capabilities in natural language understanding and generation, but it often struggle with domain-specific reasoning and factual consistency due to lack of training. This paper gives a Knowledge Graph-Based Chatbot (KG-QA Chatbot) that integrates knowledge graph retrieval with LLM-based reasoning to answer domain-specific questions accurately. Experimental results shows that our KG-QA chatbot effectively retrieves and synthesizes knowledge, achieving an accuracy of 68%, which improves the reliability of answers compared to standalone LLMs. Specifically, our model achieves precision score of 65.2, Recall score of 68.3, and then F1 score is 66.7 on Document 1, while on Document 2, it attains 67.8 precision, 71.1 Recall, and 69.4 F1 score. The results proves the considerable effectiveness of our KGQA approach in structured knowledge retrieval and question answering.