Linking Visual Concept Detection with Viewer Demographics

Adrian Ulges, Markus Koch, Damian Borth
ICMR 2012, Hong Kong, Hong Kong, ACM, ACM, 6/2012
Best Paper Award


Abstract:

The estimation of demographic target groups for web videos with applications in ad targeting ­ poses a challenging problem, as the textual description and view statistics available for many clips is extremely sparse. Therefore, the goal of this paper is to link a clip's popularity across different viewer ages and genders on the one hand with the video content on the other: Employing user comments and user profiles on YouTube, we show that there is a strong correlation between demographic target groups and semantic concepts appearing in the video (like "teenage male" and "skateboarding"). Based on this observation, we suggest two approaches: First, the demographic target group of a clip is predicted automatically via a content-based concept detection. Second, should sufficient view statistics already give a good impression of a video's audience, we show that this information can serve as a valuable additional signal to disambiguate concept detection. Our experimental results on a dataset of 14,000 YouTube clips commented by 1 mio. users show that ­ though content-based viewership estimation is a challenging problem ­ suitable demographic groups can be suggested by concept detection. Also, a combination with demographic information as an additional signal leads to relative improvements of concept detection accuracy by 47%.

Files:

  http://www.icmr2012.org/
  ICMR2012-LinkingConceptDetectionWithDemographics.pdf

BibTex:

@inproceedings{ ULGE2012,
	Title = {Linking Visual Concept Detection with Viewer Demographics },
	Author = {Adrian Ulges and Markus Koch and Damian Borth},
	BookTitle = {ICMR 2012},
	Note = {Best Paper Award},
	Month = {6},
	Year = {2012},
	Publisher = {ACM},
	Organization = {ACM}
}

     
Last modified:: 30.08.2016