Research On The Internet Public Opinion Based On Formal Concept Analysis

In recent years, with the rapid development of the Internet, the number of netizensincreases rapidly. The information on the Internet is abundant and complex. It makes theInternet become an important platform for public opinion emergence and transmission. Ourcoutry is experiencing a critical period which is filled with all kinds of problems. The Internetpublic opinion has become one of these focuses of the government and the social. In order toreinforce the network management and make timely and correct navigation to public opinion,accurately monitoring and objectively analyzing Internet public opinion has become prettyurgent for governments. Consequently, it is urgent and crucial to detect and track Internetpublic opinion by harnessing the power of modern computer technologies. It has importantpractical significance and academic value for building a harmonious society.The paper proposes a method that makes a study on Internet public opinion with FormalConcept Analysis (FCA). The main job is summarized as follows. First, collect the datasetfrom the Internet. Second, select terms with TF*PDF weighting scheme. Third, find topicswith FCA. A concept in FCA is used to express a topic and the intent of the concept is used torepresent topic content. The experimental result shows that our approach is feasible for topicdetection.The paper also proposes a linear regression based news topic hotness calculationapproach. Media focus, user attention and timeliness are considered as news features. Mediafocus, user attention and timeliness are independent variables while news topic hotness isdependent variable to confirm multiple linear regression equation. This method is validatedby correlation coefficient test (r test), goodness of fit test (R2test), regression coefficientsignificance test（T test）and regression equation significance test (F test).