Research On The Technology Of Grain Network Public Opinion Analysis

As the Internet is developing, the quantity of Internet users expanded, the network is not just a place of information resources, and it has become a space for people to express opinions. For network hot popular event of grain, people browse for the latest news on network platform, express and show their emotion, attitude and opinion. Network news has proved to be an important symbol to reflect grain events. Under this background, it is very important that how to make rational use of the network platform to enhance the monitoring of grain network public opinion, accurately and seasonable find hotspot of grain, then track the hot event and make the warning. It is the key for the government and relevant organization to grasp grain events; it also is the premise and foundation to provide decision of the grain crisis.In the paper, comprehensively elaborated research prospects and recently research on international of the grain network public opinion, comprehensively analyzed the commonly technology of grain public opinion. The mainly technologies including information acquisition, Chinese participle, feature weight calculation and selection, hot topic detection, hot tracking technique and so on. Based on the above theory, this paper made further research on the following contents.(1) Aiming at the deficiency of Chinese segmentation method which based on dictionary, this article set about the maximum matching algorithm to backtrack, added the detection method for disambiguating words which the chain length of 2, simultaneously put forward the combination type segmentation technique based on term frequency. The experiments indicated that the algorithm in this paper not only optimizing the detection of unknown words, but also improving accuracy of word segmentation.(2) Based on the commonly text similarity calculation formula and the characteristics of the dynamic development on grain network public opinion, this paper put forward a method based on the similarity of curve to find public opinion hotspot. This method would be page views of hot issues as a factor into consideration so that it can accord with the trend of dynamic development on Internet information. After validation of this paper, the improved method could realize the target of hot topic mining.(3) According to the investigation above, under the environment of MyEclipse used the programming language of Java, this paper realized the function of grain network public opinion system, mainly including hot topic detection and tracking, subject matter of automatic summary.