2013 IEEE 13th International Conference on Data Mining Workshops (2012)
Brussels, Belgium Belgium
Dec. 10, 2012 to Dec. 10, 2012
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICDMW.2012.28
This paper proposes a method for analyzing social interests. The most important characteristic of the proposed method is that its analysis is not disturbed by stealth marketing. Since conventional methods analyze social interests using posters' behavior, stealth marketers can mislead conventional methods by posting unreliable information. To solve this problem, the proposed method analyzes viewers' behavior. DNS query logs are the information source used to analyze viewers' behavior. By counting the number of viewers for each website, the proposed method tries to find current issues that attract social interests. The second characteristic of the proposed method is the use of cardinality. Cardinality is the number of unique viewers for each WWW page. Since the DNS caching mechanism hides the real access frequency of WWW pages, the proposed method uses cardinality to count the number of unique viewers. To show the usefulness of the proposed method, we have analyzed how Japanese people reacted on the eco-point system and the switchover to the digital TV broadcasts. The ability to analyze the relationship between the social interests and their changes is the third characteristic of the proposed method and is revealed by the experiments which use a three-year long country level DNS query log.
Servers, IP networks, Digital TV, Internet, Switches, Government, Predictive models, Cardinality, Social Interest, viewer, poster, DNS
Takeshi Mitamura, Kenichi Yoshida, "Viewers' Side Analysis of Social Interests", 2013 IEEE 13th International Conference on Data Mining Workshops, vol. 00, no. , pp. 301-308, 2012, doi:10.1109/ICDMW.2012.28