Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
SOE: Fachverband Physik sozio-ökonomischer Systeme
SOE 18: Social Networks
SOE 18.3: Vortrag
Donnerstag, 19. März 2015, 11:45–12:00, MA 001
Prediction of topics' survival using large-scale social data: case of comedian popularity — •Kenta Yamada1,2, Ryo Tamaoka1, and Kiyoshi Izumi1,3 — 1Graduate School of Engineering, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656 Japan — 2PRESTO, Japan Science and Technology Agency, 4-1-8 Honcho, Kawaguchi, Saitama, 332-0012, Japan — 3CREST, Japan Science and Technology Agency, 4-1-8 Honcho, Kawaguchi, Saitama, 332-0012, Japan
We proposed a new indicator analyzing large-scale textual data in blogs, which predicts future popularity of a topic after a related event. The indicator was tested about the prediction of comedians' popularity after famous comedican contests on television. There are some popular comedian contests in Japan such as the M-1 grand prix. We can universally observe clear peak and power law decaying in the number of blog entries including a champion and vice-champion name (34 samples) after contests as well as the cases in which the number of entries including the event's name follows power function after the events such as Christmas in the previous study [1]. We fitted the number of entries including the comedian's name using five days data after the contest by a power function and cumulated differences between fitted line and actual data from 6 to 12 days after the contest. We found that this index of cumulative differences has a good predictive capability for the number of future (11months later) entries about the comedian.
[1] Y. Sano, K. Yamada, H. Watanabe, H. Takayasu, and M. Takayasu, PRE 87, 012805 (2013).