Research On Collection, Analysis And Visualization Of Micro-blog

With the development of social media, such as micro-blog and wechat, these new inventions not only has changed people’s ways of life, but also has brought the huge volume, multiple dimensions, unstructured data at the same time. Most researchers believe that these data are the treasures of this era, and it makes data oriented scientific research more and more popular. This paper discusses the micro-blog data oriented research work from three aspects: the first is micro-blog data collection, and the second is the methods to find new emotional words based on micro-blog data, the last is visualization research based on the micro-blog forwarding data.(1) For collection of micro-blog data, this paper first analyzes two different ways to simulate login authentications, and respectively discusses the advantages and disadvantages of the two methods. Secondly, after acquiring authentication, this paper introduces four types of data collection:the users’personal information, micro-blog information, users’like list and forwarding micro-blogs and comments of a single micro-blog. It is the foundation corpus for follow-up study.(2) Due to its characteristics of informal conversation, micro-blog data contains a large number of new emotional words which has not been recorded in related database. Based on the micro-blog data, this paper does some research on word-level sentiment analysis. First of all, this paper uses statistic methods to identify new words in micro-blog data. Secondly, this paper uses distributed representations of words, trained on micro-blog data using neural network model, to capture the semantic and syntactical information among words. At last, this paper proposes methods for new emotional words discovery based on distributed representations of words. The result demonstrates that the methods used in this paper has some practical values.(3) Based on the forwarding data of micro-blog, this paper conducts visualization analysis for the forwarding process of a single micro-blog. First, this paper uses the micro-blog forwarding data to build spreading network. Then according to the personal information data, this paper analyzes the implementation of visualization on three aspects:the selection of nodes, hierarchical information display and the interactive function designing. The visualization is really useful to find the important nodes and judge the influence of micro-blog dissemination...