Research And Implementation Of Correlation Analysis For Microbloging Influencers And Topic Popularity

As an important product of information technology development, with its openness, convenience and other features, microblogging has made it possible for people to communicate with each other in broader fields, and accomplished the process from generation to popularization, and finally transformed to an important public opinion platform in a very short period of time. Social events originated in reality are now spreading out on the microblogging platform, from the embryonic stage at the very beginning to the expanding stage via the transformation by opinion leaders with all kinds of viewpoint collisions, which often leads to the evolution of the topic content. In the process of the evolution stated above, it is the prerequisites for the in-depth research and analysis of the topic context that how to quantify the impact of the spread of a microblog and how to discover opinion leaders and measure their contribution in the information dissemination accurately.In this paper, conduct topic analyses based on the existing studies in both macro and micro perspectives: first, study in macroscopic view "What sequential association exists between the topic-related opinion leaders and the topics" and "Which opinion leaders have impacts on the topic evolution" and other issues. On this basis, study in microscopic view "Different types of bloggers published what the main characteristics of the microblogs " and "What impact will be made to the topics by the microblogs with different content characteristics" and other issues. take both experimental and empirical approaches to conduct experiments and data analysis based on microblog topic examples in reality. Related work and innovations are as follows:1) Collect the microblog data on the topic of "Attack on Nanjing Hospital nurse", analyse the study feasibility of the data based on facts, discover the topic opinion leaders through a common and easy method based on limited user relationships on a microblog topic; and also establish a model for "opinion leaders’ influence on a certain topic" and a model for "topic heat index", providing support for the follow-up work.2) study and implement a technique for microblogger influence measurement based on dynamic time warping(DTW) distance. In macroscopic view, measure the distance between the "opinio n leaders’ influence on a certain topic" and the "sequential timing of the topic heat index" distinguishing between different time windows and time granularity, find an association between the ones with smaller distance and that some of the opinion leaders play a key role in backstage promotion, guidance and control for the microblog topic evolution in the macro perspective.3)Study and implement a critical microblog discovery technique based on influence saturation. In microscopic view, analyse the microblog content characteristics sorted by both "forwarding count" and "influence saturation", study the sequential characteristics of the "influence saturation" based on user forwarding behavior and influence coverage, propose a model for the evaluation of single microblog influence saturation. O n this basis, find by experiments that microblog content and bloggers type sorted by "forwarding number" is rather different from that sorted by "influence saturation" and make a quantitative analysis of their different roles in the topic.In summary, in this paper primarily study the correlation between the opinion leaders and the microblog topics based on DTW distance via the Internet topics originated in reality like "Attack on Nanjing Hospital nurse", analyse the topic evolution in macroscopic view; and introduce the concept of "influence saturation", study the microblog topic structures from the perspective of "forwarding number" and the perspective of "microblog influence saturation", which provides a useful compleme nt to the existing topic analysis methods.