Font Size: a A A

An Online Video Advertising System Based On Users’ Tags

Posted on:2013-01-03Degree:MasterType:Thesis
Country:ChinaCandidate:J YeFull Text:PDF
GTID:2248330392457781Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the rapid development of online video industry, the revenue of online video ad-vertising has been maintained a good momentm of growth. Existing internet video adver-tisement delivery solutions deliver ads randomly, so that ads are rarely related to videocontent. In addition to this, the fixed ads delivery timing leads to poor user experience,which greatly affects the ads effect. Using C/S mode to deliver ads, current video ads sys-tem occupy great bandwidth of advertising source server.CMAd, an online video advertising system which based on users’ tags, can effectivelyenhance the result of advertising, and reduce the pressure of the ad source servers duringthe advertising distribution. CMAd generates the hot spot and descriptions of the videobased on users’ tags, and then put in the ad whose content is most relative with the hotspot’s decriptions belongs to the video. As the users’ tags reflect the contents of the videobeing played, CMAd can effectively increase the correlation between the ads and the video.CMAd proposes an advertising time chosing strategy based on users’ attention on the video.Given an online video with hotspots generated by users’ comments, CMAd is able to detectad insertion points based on user’s attention on the video, compute relevance between eachpair of insertion points and ads, and select the most relevant ad for each insertion points.CMAd use the non-structure of the topological organization; dynamically update the node’stopology information according to the cache overlay and the network delay between thenode and its neighbors. Finally, CMAd uses the node’s neighbor auxiliary service to theadvertising data request.According to log of CMAd system, the similarity of ads and the video being played arehigher than75%. Less than10%of users would close advertisement windows and theclick-through rate is as high as0.73%. The bandwidth consumption is decreased by70%compared to C/S mode based ads delivery solutions.
Keywords/Search Tags:Online Video Adversiting, Contextual Relevance, Attention on Video, DataScheduling
PDF Full Text Request
Related items