Research Of Identifying Methods About The Network Opinion Leaders Based On Semantics

Since Chinese has the characteristic of flexible application and complex ideograms in the communication environment of computer medium, in the opinion leader’s identification progress, besides the post’s reply structure and social network attribute of related nodes, text content and its emotional tendency should paid more attention. Currently the identification and discovery of opinion leader has given highly attention and has becoming a research hotspot. Pointing at the problem that exist in current network opinion leader recognize method, on the basis of consequence diffuse model-IDM, combined with the tendency analysis, this paper put forward an influence diffusion model based on semantic comprehension, and applied it into the network opinion leaders’ discovery methods research. The thesis completes the following tasks:(1)Analysis and discussion of emphasis and limitations of the existing network opinion leaders’ identify method.The anonymous and virtual performance of computer mediated communication environment leads the problems that there are fuzziness and indeterminacy in the identification progress of network opinion leaders. Take consequence diffuse model-IDM and related improvement for example, this paper discuss the existing choke point and deficiency in the opinion leader reorganization method, put forward the method that apply text analysis into consequence diffuse computing.(2)Modeling the consequence diffuse model based on semantic tendencyOn the basis of consequence diffuse model-IDM, a consequence diffuse computing method which combines inclination analysis is proposed. Abstracting the emotion inclined in reply relationship as the support to the main article, and then makes an objective quantify analysis on consequence diffusion.(3) Identification method is proposed based on the semantic network opinion leaders.In view of semantic sentiment consequence diffuse model, this paper put forward a network opinion leaders’ identification method based on network structure and semantic sentiment, which can identify and filter opinion leaders automatically and finally verify the availability and accuracy by experiments.