In this information age, Social Networking Services contribute a significant amount of contents in creating a knowledge based society. Nowadays, there are more than 500 million tweets sent per day in Twitter. Such drastic growth of contents brings new opportunities for human beings to discover their surroundings more effectively in a timely manner. Moreover, these types of services evolve not only in a perspective of scalability, but also in the view of indicating more meaningful information regarding what happens in the world. Numerous news agencies are broadcasting breaking news via Twitter and people would like to leave comments with their own opinions as well. However, there are differences between events that news media are more willing to cover and news stories that people are more interested in. Furthermore, as people are becoming the largest sensor network, trending topics are not only led by media, but also by the public, and hence it is worth pondering how they affect each other. In this paper, we focus on studying these concerns by building a system, Ushio, analyzing Twitter streams in both the tweets updated by multiple news agencies and those appearing in the public timeline. We describe our design and implementation of this system, which extracts named entities from the Twitter streams and generates corresponding statistics with its relational model. We then show how we use these data to find trending topics and real focus from both media and the public, as well as discover their related topics along with the correlation indicating the leading role between them for assorted topics.