Twitter Estimation of alse Rumor Diffusion Model and Prevention Model of alse Rumor Diffusion on Twitter 5 6 7 7,8 Takashi Shirai, Takeshi Sakaki, ujio Toriumi, Kosuke Shinoda, Kazuhiro Kazama 5, Itsuki Noda 6, Masayuki Numao 7 and Satoshi Kurihara 7,8 Graduate School of Information Science and Technology, Osaka University Tokyo University Nagoya University RIKEN 5 NTT 5 Network Innovation Laboratories 6 6 National Institute of Advanced Industrial Science and Technology 7 7 The Institute of Scientific and Industrial Research, Osaka University 8 8 JST CREST Abstract: Twitter is a famous social networking service and has received attention recently. Twitter user have increased rapidly, and many users exchange information. When Tohoku earthquake and tsunami happened, people were able to obtain information from social networking service. Though Twitter played the important role, one of the problem of Twitter, a false rumor diffusion, was pointed out. In this research, we focus on a false rumor diffusion. We propose a information diffusion model based on SIR model, and discuss how to prevent a false rumor diffusion. [] SNS 567-7 8- E-mail: shirai@ai.sanken.osaka-u.ac.jp acebook Twitter SNS Twitter http:www.facebook.com http:twitter.com
[9]SNS SNS SNS Twitter SNS SIR Twitter SNS Twitter Twitter Twitter Huberman Twitter @ [5] Akioka Twitter []Akioka Twitter Twitter Twitter Kawk Twitter [7] Kawak Twitter [] Weng Twitter PageRank Twitter TwitterRank [] Balshy URL URL [] Twitter Castillo [] [] Twitter Castillo. SIR SIR Kermack [6][8] SIR S Suseptible I Infectious R Recovered S I ρ (S I) I I
ρ (I R) R t S S(t) I I(t) R R(t) S I R ds(t) = ρ (S I) I(t)S(t) di(t) = ρ (S I) I(t)S(t) ρ (I R) I(t) dr(t) = ρ (I R) I(t) () N N = S(t) + I(t) + R(t) SIR S I S S I R I R S I R SIR S I ρ (S I) S I I S S I get I S I R I get R get. SIR = R I R I R SIR S I R.. SNS SNS S I get I R get R R get R S I get I ρ (S I) S I ρ (Iget I) ρ (S R) ρ (Iget R) ρ (I R) ρ (Rget R) N Twitter t S I get I R get R S(t) I get (t) I(t) R get (t) R(t) SNS SIR S SNS
I t S S(t) N I get Iget(t) N ds(t) = N I(t)S(t) di get(t) = ( ρ (S I) ) N I(t)S(t) ρ (Iget I) N I get(t)i(t) di(t) = ρ (S I) N I(t)S(t) +ρ (Iget I) N I get(t)i(t) () ds(t) = N I(t)S(t) N R(t)S(t) di get(t) = ( ρ (S I) ) N I(t)S(t) ρ (Iget I) N I get(t)i(t) di(t) dr get (t) dr(t) N I get(t)r(t) = ρ (S I) N I(t)S(t) +ρ (Iget I) N I get(t)i(t) N I(t)R(t) = ( ρ (S R) ) N R(t)S(t) +( ρ (Iget R)) N I get(t)r(t) +( ρ (I R) ) N I(t)R(t) ρ (Rget R) N R get(t)r(t) = ρ (S R) N R(t)S(t) +ρ (Iget R) N I get(t)r(t) +ρ (I R) N I(t)R(t) +ρ (Rget R) N Rget(t)R(t).. () SNS Twitter i : Step Step t = I Step t = R Step t = 5 : 5, =, = =.5 =5. =.5 =.5 ρ (S I) =.5 ρ (Iget I) =.5 ρ (S R) =. ρ (Iget R) =. ρ (I R) =.5 ρ (Rget R) = j i j SNS Twitter [][]. SNS SNS Twitter Twitter []
6 5 数ド ーノ S Iget I Rget R 5 6 7 8 9 5 6 7 8 9 5 6 7 8 9 シミュレーションステップ : : Step Step Step. Step RT Step5 Step6 Step7 Twitter Web LP 9,65 5,88 5
: : A- A- A- 5 5 5 : : : : 5 : 6 : 7 : 8 : 9 : : : : : : 日付 時刻ツイート数デマツイートの数訂正ツイートの数 : :. ρ (I R) 5 6 N I N R N (I R) N (I R) N I 5 5 5 5 9 : : : : 7 : : : 6 : 9 : : : : 7 : : : 6 : 9 : : : : 時間経過ユーザー数 S I R : 5: Step Step ρ (I R) 7 i i 5 5 6 8 I
6: N I N R N (I R) N (I R) N I 9,75 7,8,6.7 8: I R R I.6 5.9. 7: Step 5 Step Step I R I R R I R R I Twitter SNS i i i Twitter Twitter 5 8 5 6 9.6%.% 5 5. Twitter I get R get I get I 57.6% Twitter I get R get 9 5 S I R S I R
8 6 数ド ーノ 8 6 頻度 5 5 5 5 5 5 55 6 65 7 75 8 85 9 95 デマを受け取った回数 : 数ドー : ノンョシーレュミシ 6895 59 95 9 955 97 985 シミュレーション :S シミュレーション :I シミュレーション :R 実データ :S 実データ :I 実データ :R 6 8 6 8 6 8 時間経過 5 5 5 5: 5 数ーザー : ユターデ実 9: ρ (S I) =.5 ρ (Iget I) = ρ (S R) =. ρ (S I) ρ (Iget R) = ρ (I R) =.7 ρ (Rget R) = 5. A B C Step: R B (5 5 ) R Step: 6 A I R ρ (Iget R) 6 Twitter B C A B C C 6 SNS Twitter SNS
5 数ドーノの R 通常ルール起点ユーザー選択ルール A 起点ユーザー選択ルール B 起点ユーザー選択ルール C 8 9 5 6 7 8 9 シミュレーションステップ 6: :R SIR Twitter Twitter SNS Twitter [] Sayaka Akioka, Norikazu Kato, Yoichi Muraoka, Hayato Yamana Cross-media Impact on Twitter in Japan, Proceedings of the nd international workshop on Search and mining user-generated contents, pp.-8,. [] Yasuyoshi Aosaki, Taro Sugihara, Katsuhiro Umemoto Examining the Trend toward a Service Economy in Information Media through Changes to Technology: Influence of Twitter on Media Companies, Proceedings of Technology Management for Global Economic Growth (PICMET), pp.-5,. [] Eytan Bakshy, Jake M. Hofman, Winter A. Mason, Duncan J. Watts Everyone s an Influencer: Quantifying Influence on Twitter, Proceedings of the fourth ACM international conference on Web search and data mining, pp.65-7,. [] Carlos Castillo, Marcelo Mendoza, Barbara Poblete Information Credibility on Twitter, Proceedings of the th international conference on World wide web, pp.675-68.. [5] Bernardo A. Huberman, Daniel M. Romero, ang Wu Social networks that matter: Twitter under the microscope, irst Monday, Vol., No., 9. [6] W. O. Kermack, A. G. McKendrick A Contribution to the Mathematical Theory of Epidemics, Proceedings of the Royal Society 5A, pp.7-7, 97. [7] Haewoon Kwak, Changhyun Lee, Hosung Park, Sue Moon What is Twitter, a Social Network or a News Media?, Proceedings of the 9th international conference on World wide web, pp.59-6,. [8],,, 5. [9], http:www.soumu. go.jpjohotsusintokeiwhitepaperjahpdfindex. html,. [],,,,,, JWEIN, pp.-6,. [],,, Twitter RT,, Vol.-IAT-, No., pp.-6,. [] Jianshu Weng, Ee-Peng Lim, Jing Jiang, Qi He TwitterRank: inding Topic-sensitive Influential Twitterers, Proceedings of the third ACM international conference on Web search and data mining, pp.6-7,. [],,, 7, pp.5-,.