Dynamic Detection And Evolution Analysis Of Micro-blog Topic Based On Association Rule

Presently,Sina micro-blog is the largest micro-blog service platform in the country.Micro-blog event stream contains a large number of various areas of news events.There are a variety of topics with the topic label in Sina micro-blog.So far,there has been a lot of research on event detection in micro-blog,but few focus specifically the research of a particular topic.Some emerging hot events,originated from micro-blog,caused phenomenal influence on society.The news media and public relations team related to the event,in particular,pay extremely close attention to the development of the event.So the detection and evolution analysis of specific target topics can be in great market demand.Topics can be divided into several categories,including breaking event,most-discussed event and so on.The analysis of the existing micro-blog event detection methods finds that the current event detection methods based on micro-blog is rarely target-aware,and they are either confined to a single detection category,or unable to detect the right category of a n event.On the basis of summarizing the previous work,this paper mainly carries out research focusing on the above problems from the following aspects: firstly,we analyzed the characteristics of micro-blog content,combining with micro-blog's topic labeling and association rules mining methods,to study how to target the topic detection and tracking evolution analysis.Secondly,considering that the topic can be divided into many categories,we learned from the application of association rules in the mall customer shopping habits and analyzed the role of micro-blog in the label on the topic,we modify and put forward 4 kinds of evolution mode of association rules which including new rules,emerging rules,changing rules and dead rules based on the micro-blog topic label.Using the 4 kinds of evolution mode of association rules,we can simultaneously detect a variety of categories of topics,and clear the purpose of their respective categories which provides support for the subsequent evolution of the topic.Thirdly,the methods used in our topic detection are directly applied to evolutionary analysis,and four evolutionary patterns of association rules are used to analyze the evolution of target topics.The task of detecting and evolving the topic is solved by the same method,which reduces the complexity of the topic detection and evolution analysis link to a certain extent.Fourthly,we apply the existing methods to the micro-blog topic detection method and compare it with our method,the result analysis of each method is discussed.We can get the following points from the experimental results.Firstly,our method can effectively detect and analyze specific target topics.Secondly,the methods used in this paper can be used to detect different kinds of topics,and to define their categories.Thirdly,the method in this thesis can be applied to the analysis of micro-blog's topic detection and evolution at the same time.