Public Opinion Research Based On Network Comments

According to CNNIC 25th China Internet Development Survey: the scale of China's Internet users reached 384 million people, coverage reached 28.9%, exceeding the global average, ranking second in the world at the end of December 31, 2009.The emergence and flourishing of the Internet has greatly expanded the public space in Chinese society, Internet users can not only receive information, but can also express their interest, giving vent to emotions through the Internet media. Once public opinion appears on the Internet, its follow-up communication will inevitably have a profound impact on our lives,to find out the regularity of network public opinion can do advance warning for some problems which may lead to social stability, so that the relevant departments can take certain measures to affect the communication process. That has great significance for government's decision-making, network purification and prosperity and the construction of a harmonious society.This paper selected network news message board as the data source and chose Tencent network news commentary to be the representative data, to study public opinion on the network from a point of view of quantity. First, to draw a time-series graph on raw data, to analyze all comments curve as a whole from the number and growth trends of comments,we can get that Livelihood class information obtained the highest level of concern All news reviews have gone through the formation of high, volatile and ultimately the development of desalination,To further investigate the growth regulation of network public opinion essentially and a Dynamically to map the cumulative curves After processing Based on the original data. To cluster the curves by Using self-organizing map neural network clustering method, And do detailed analysis of the characteristics for each type of curves.Using the concluded analysis above, to mode for each type of curve by using polynomial fitting, and nonlinear fitting method. And briefly describes how to forecast the trend of new comment on the number of unexpected events and to illustrate the prediction method is good by using examples.Finally, the paper provides a recommendation for how to take measures to the Emergency for the related departments.