With the advance of social platforms, people are sharing contents in an unprecedented scale. This makes social platforms an ideal place for spreading rumors. As rumors may have negative impacts on the real world, many rumor detection techniques have been proposed. In this proposal, we summarize several works that focus on two important steps of rumor detection. The first step involves detecting controversial events from the data streams which are candidates for rumors. The aim of the second step is to find out the truth values of these events i.e. whether they are rumors or not. Although some techniques are able to achieve state-of-the-art results, they do not cope well with the streaming nature of social platforms. In addition, they usually leverage only one type of information available on social platforms such as only the posts. To overcome these limitations, we propose two research directions that emphasize on 1) detecting rumors in a progressive manner and 2) combining different types of information for better detection.