In this paper, we propose a way to transform traditional Q&As into conversational Q&As for an efficient information retrieval in special knowledge. Special knowledge involves difficult words. It requires users to raise a series of questions and get the answers to them to pinpoint the desired information. And, conversational Q&A is appropriate than the traditional Q&A because it allows a user to narrow down searches in a solution space. To transform a given set of Q&As to conversational Q&A system for special knowledge search, we first explore not only the present traditional Q&A systems and conversational Q&A systems for general knowledge search, but also those for special knowledge search. From this, we induce an appropriate search process in conversational Q&A systems for special knowledge. Secondly, we build an ontology with the help of machine learning to support the navigation in special knowledge. Finally, we give a way to evaluate performance after embedding the ontology on our search process of conversational Q&A. We apply this procedure to the case of Korean simplified taxation in a Korean Q&A system, Naver Jisik-In Q&A. We found that searching through Jisik-In Q&A with ontology has better usability than using Jisik-In Q&A only. Therefore, this study aims to improve the usability of special knowledge search, lower the threshold of special knowledge, and develop special knowledge as general as common knowledge using conversational Q&A based on ontology. However, as the number of user experimented is limited and the classifier for the extracted words from existing Q&A system should be reviewed by tax expert, so the future work is demanded.