Finding a good research topic is very crucial for a researcher, especially a new researcher, such as PhD student. Generally, a PhD student’s research interest in the beginning is not clear, and he have to spend much time reading a lot of research literatures, step by step, his research preference will become clearer and clearer, and easier to find an advanced research topic in which he is really interested. So, reading a certain number of research literatures is critical for a PhD student to find final research topic. But, which literature should be selected to read? How to quickly clear user preference and determine final research topic by reading fewest literatures? As far as we known, there is no system used to support the user to find research topic. In the paper, a research topic decision support system is originally proposed and creatively developed to help a user to quickly find potential research topic based on iteratively paper recommendation. Starting from user’s initial unclear research preference, through continuous recommending appropriate literatures to the user, then the user read and critiques them by rating until he find final research topic. During the process, since the user read the most appropriate literature in each round, he can quickly find the research topic while spending minimal time.