Key Technologies Research On Detection And Control Of Internet Public Opinion

With the growing number of netizen, more and more netizens are keen on forum, blog, microblog and other network media as a place where is the most concentrated public opinion expression, the most free-flowing dissemination of information. Detection and control of internet public opinion related to several key technologies, and academic research focus on information collection and extraction, clustering and analysis of topic, and control public opinion.In paper there will be 109 colleges Forum for the study, it will have a detailed discussion on three key technologies that is related to monitor and intervene to internet public opinion: incremental crawling, information extraction, and control public opinion.Incremental crawling divides acquisition process into two parts: the off-line part and on-line part. In the off-line part, first of all, In the off-line part, first of all, through the collection of sample pages we have get the crawling path that the crawler has to pass and page attribute information; and then, through the identification of key resources, it has been able to extract their path. In addition, speaking time and effective information content have been applied to the Poisson model which based on the time model to get reasonable time by incremental crawler. In the on-line part, it will use the result of the offline part to guide the collector work, greatly improve the accuracy of the information acquisition and reduce the burden on the collector.In the part of information extraction, through the use of web structure similarity which exists in the post, by combined with the feature of User-Created Content, we propose algorithms that locate title subject and reply to messages by the index path and remove intensity noise which is in reply to message that may exist interference.In the part of control public opinion, it makes full use of operation-interface which is provided by the forum, through the discussion with the problems like identifying code hidden identity and so on; finally we can end up with a unified solution for different forums that need to implement intervention in public opinion.