Analysis of Retweet under the Great East Japan Earthquake

Transcription

1 1,a) Twitter 4 Twitter Twitter, Twitter,, Analysis of Retweet under the Great East Japan Earthquake Abstract: In this paper, we analyzed the 400 millions of Tweet data which posted around the Great East Japan Earthquake to find how the twitter used and how the Twitter was influenced by the disaster. We modeled the time series data of Retweet by Log Normal Mixture Model. By analyzing the model, we found that the peak times of the retweets are become shorter, and there are few long range retweets after the disaster. As a result, we can say that the role of the Twitter was changed from communication tools to information sharing tools since the Great East Japan Earthquake were occurred. Keywords: Social Media, twitter, the Great East Japan Earthquake, Time Series Analysis, Log-Normal Mixture Model 1. WEB Twitter Twitter Java [3] 1 the University of Tokyo 2 RIKEN 3 NTT NTT Network Innovation Laboratories 4 Osaka University 5 National Institute of Advanced Industrial Science and Technology a) tori@sys.t.u-tokyo.ac.jp Twitter Twitter Kwak [4] Twitter Twitter Twitter 2010 Twitter Mendoza [5] Heverin [1] Twitter c 2012 Information Processing Society of Japan 1

2 Twitter [8] [6] [7] URL [2] Twitter Twitter Twitter TwitterAPI ( 1 ) 200 ( 2 ) 200 ( 3 ) TwitterAPI ,6, :00: :59:59 362,435, Rate of Retweet 1.8% 18% Avg. Retweet Twitter c 2012 Information Processing Society of Japan 2

3 RT *1 Twitter RT RT RT RT RT 3 RT RT *1 ITmedia /12/news013.html (7-10 ) (12-15 ) (17-20 ) ( 1 ) 10% ( 2 ) 95% c 2012 Information Processing Society of Japan 3

4 4 p(x) = K w k P k (x i µ k, σk) 2 (2) k=1 w k µ k, σ 2 k EM x = {x 1, x 2,, x N } EM E-Step: µ, σ ω x i k z ik = ω kf(x i µ k, σ 2 k ) K l=1 ω lf(x i µ l, σ 2 l ) (3) Q(x) (Gausian Mixture Model) x f(x) = 1 (ln x µ)2 e 2σ 2 (1) 2πσx x Q(x) = N K i=1 k=1 z (t) ik log ω kf(x i µ k, σ 2 k) (4) M-Step: µ k, σ k, ω k µ = i=1 z ik ln x i i=1 z ik (5) σ = i=1 z ik(ln x i µ k ) 2 i=1 z ik (6) ω = 1 N z ik N (7) i=1 E-Step M-Step µ, σ, ω 100 K (AIC) c 2012 Information Processing Society of Japan 4

5 AIC 7 (3/7-10) (3/12-15) (3/17-20) ω?? Twitter Twitter 1 2 Twitter Twitter ( ) NTT c 2012 Information Processing Society of Japan 5

6 [1] Thomas Heverin and Lisl Zach. Microblogging for Crisis Communication: Examination of Twitter Use in Response to a 2009 Violent Crisis in Seattle-Tacoma, Washington Area. In Proceedings of the 7th International ISCRAM Conference, Seatle, Washington, [2] Takeru Inoue, Fujio Toriumi, Yasuyuki Shirai, and Shinichi Minato. Great east japan earthquake viewed from a url shortener. In Proceedings of the Special Workshop on Internet and Disasters, SWID 11, pp. 8:1 8:8, New York, NY, USA, ACM. [3] A. Java, X. Song, T. Finin, and B. Tseng. Why we twitter: understanding microblogging usage and communities. In Proceedings of the 9th WebKDD and 1st SNA-KDD 2007 workshop on Web mining and social network analysis, pp ACM, [4] H. Kwak, C. Lee, H. Park, and S. Moon. What is Twitter, a social network or a news media? In Proceedings of the 19th international conference on World wide web, pp ACM, [5] Marcelo Mendoza, Barbara Poblete, and Carlos Castillo. Twitter under crisis: can we trust what we RT? In Proceedings of the First Workshop on Social Media Analytics - SOMA 10, pp , New York, New York, USA, July ACM Press. [6],,,. Twitter RT , [7],,. Twitter , [8]. 23. CIAJ journal, Vol. 51, No. 10, pp , c 2012 Information Processing Society of Japan 6