Research On Analysis Of Network Public Opinion Tendency Based On Korean

With the rapid development of Internet,more and more people get used to getting information from Internet.Users have access to all aspects of information such as finance,politics,sports,entertainment and so on.So people can expound their own opinions and views about social top events.As a consequence,those opinions and views formed the Internet public opinion.In another way to explain Internet public opinion is that public opinion is on the basis of the public affairs combined people’s thoughts with results of the events on the Internet.Network has the character of openness and virtualization and it’s lack of restraint of legal and moral,so it’s easy to generate disharmony speech on it.Thus the importance of Internet public opinion tendency analysis is particularly prominent.This paper did research into Korean public opinion tendency analysis based on Chinese public opinion tendency analysis.After building the Korean sentiment dictionary,this paper applied it to do feature selection and compare its performance with other feature selection algorithms.Thus we can analyze Korean news texts’ emotion tendency to judgment their polarities are either positive or negative.The main contents of this paper can be summarized as following respects:1.Connected with Baidu API,translate a Chinese dictionary into Korean word library.Construct a Korean word library artificially by querying open resources of Korean dictionary and relevant information.2.When building a data set,translate Chinese news and Korean news with Baidu translator(conneted with Baidu API)to each other,construct a Korean sentiment dictionary by using PMI algorithm based on Chinese sentiment dictionary of How Net..3.Modify Paoding tokenizer,and add custom dictionary to it’s word library,to make sure it can complete Korean word segmentation.4.Do feature selection by combining sentiment dictionary with MI algorithm after text preprocessing,and calculate weight for the features which meet the conditions.5.Judge texts’ tendencies based on the SVM classifier,conclude the polarities of the texts,and get the positive value and negative value.