Popularity of bloggers and the amount of information in the blogosphere increase fast. Blogs have constituted a dynamic and tightly social network by using frequent links and information interaction, and become an important source of information for the real world. Most researches on blog mainly concentrate on blog definition and identification, content mining, community discovery, importance analysis, blog search and spam blog identification. Methods and technologies of link analysis and natural language processing are used in most works, and some blog-specific methods are proposed. This paper analyzes and compares these researches on blogosphere. Problems of current topics are discussed, and finally future directions are proposed in this paper.