Research On Ontology-based Internet Public Opinion Mining

As the growth of Internet,network quickly spread,andthe number of Internet usersgrows rapidly.Thus, the network public opinion gradually become thereal-timebarometer mapping social public opinion. To some extent, it can be the mostdirectand rapid reflection of the status and development trend of economic,political andculture. Naturally,the network public opinion analysis has becoming one of the keytechnologies of understanding the society and public opinion.However, regulators tendto be more concerned about a particular area or level of public opinion situation.Butfindinga local information associated withspecific domain from the mass of searchresults based on keywordsis very difficult. On the other hand, the Internet’sopenness, sharing, flat makesnetwork public opinion and supervision no longer to beregional and noboundary.Therefore, when searching correct expression of the expected,the lack of semantic understanding of search keywords totally ragged.According to the research purpose of the demand, the research work of thispaper should be from the following contents:Understand the role of the body in the course of public opinion analysis.The publicopinion analysis will based on the semantic;This paper presents a Web-based ontologymining for the lack of semantic understanding in traitional web mining.Beforeexcavation analysis, use the semantic to understand the body of the text. This will beapplied to the body of knowledge among Web mining.Making research in both areascan promote each other to solve this complex social problem of public opinion analysis.Research the text classification algorithm combined with domain ontology, andimprove it to accommodate the special needs of public opinion analysis.Meanwhile, thetraditional classification methods are not well adapted to the special needs of publicopinion analysis.This paper will improve the text classification algorithms to the suit thespecific needs of public opinion analysis.Text classification will lead to unevendistribution of sample bias, public opinion analysis. This improvement reduces theeffective sample classification results on uneven, thereby reducing the deviation.Improved sentimentof traditional algorithm, make the calculatingfor sentimentmore accurate and effective;Finally, this papermakes the quantitativecalculation of the emotional tendentious more completly.It makes the analysis moreaccurate, is very important for public opinion analysis.