| In recent years, with the rapidly development of micro-blog network and the expandingof user group, the micro-blog network has been paid close attention by more and more peoplefor its public opinion monitoring and advertisement delivering. At present, the micro-blognetwork has been the hot spot of social network research. In this paper, based on the domesticand foreign research, it is aimed at making a more accurate evaluation of the real influence ofmicro-blog information and finding a more rapid and efficient way to pick out the initial usergroup with the maximal influence for the current evaluation is not accurate enough and thecharacters of micro-blog network are ignored in traditional influence maximization algorithm.In current research, there are few studies on micro-blog network information influenceevaluation. In the field of public opinion, we could discover hot news and realize monitoringthrough the evaluation; in business scale, businesses could see the advertisement deliveringeffect by accurate evaluation. However, the existing evaluating method could not be soaccurate now for it is made only by the amount of forwarding and comments. A user qualityoriented information influence evaluating algorithm is proposed in this paper which focuseson the quality of users participate in the information propagation. It judges informationcomprehensively by adding repetition avoiding mechanism and punishment mechanism.Finally, it is proved by experiment that could screen inauthentic information influence broughtby fake advertisement and robot users to accurate information influence.Influence maximization of social network is always the study hot spot. And the influencemaximization study for micro-blog network could not only prevent the large-scale diffusion offake information but also could be used in advertisement delivering. Traditional studies aremade mainly aimed at the data set provided by other media. Then the influence maximizationalgorithm based on the users’ behavior proposed in the paper takes the characters ofmicro-blog, such as high dynamic, large data size, etc. into consideration. This algorithmemphasized on the users’ activity and behavior bias and improved the propagation probabilityof Independent Cascade Model. At last, through the experiment compare of data set ofTencent micro-blog network and artificial network data set for KDDCUP2012, it turned outthat there was great improvement than those heuristic algorithms on influence maximization effect. It is much better than greedy algorithm on time complexity and could screen theimpact of robot users and zombie users effectively.Above all, in this paper, it made a research on the micro-blog network informationimpact assessment and influence maximization focusing on users’ attributes and behaviorunder the condition of the existence of a lot of zombie users and robot users in the micro-blognetwork. And the feasibility and actual result has been verified by experiment which madecontribution. |
PDF Full Text Request