[1] HITS EigenRumor Web PageRank 情報の要求 投稿者推薦システム 投稿者の重要度推定 ( 本研究 ) の引用回数から推定 投稿者のネットワークから推定 個人的な興味を考慮した部分 1 投稿者のランキング Web EigenRumor Kri

Transcription

1 DEIM Forum 21 C A calculation method of blogger s importance using influences to others in micro-blogs Kazuki YOSHIMOTO, Yu SUZUKI, and Masatoshi YOSHIKAWA Undergraduate School of Informatics and Mathematical Science, Faculty of Engineering, Kyoto University Yoshida-Honmachi, Sakyo, Kyoto, Japan Graduate School of Informatics, Kyoto University Yoshida-Honmachi, Sakyo, Kyoto, Japan spzuka@db.soc.i.kyoto-u.ac.jp, {ysuzuki,yoshikawa}@i.kyoto-u.ac.jp Abstract In this paper, we propose a novel method for assessing quality of twitter users using the number of retweets and followers. We have two assumptions that if a user submits messages which are important for the other users, these messages are frequently retwitted by the other users. Another assumption is that if a user submits important messages frequently, the user is followed by the other users. Moreover, if a qualified user retwits or follows a user, the quality score should be increased higher than the case of an unqualified user. We propose a quality score calculation method based on these two assumptions. In our experiments, we confirmed that our proposed method can calculate quality scores with high accuracy. Key words Micro-blog, Blogger network, Information recommendation 1. CGM(Consumer Generated Media: )

2 [1] HITS EigenRumor Web PageRank 情報の要求 投稿者推薦システム 投稿者の重要度推定 ( 本研究 ) の引用回数から推定 投稿者のネットワークから推定 個人的な興味を考慮した部分 1 投稿者のランキング Web EigenRumor Kritikopoulos [2] [3] Agarwal [4] 2. 2 Twitter Bernardo [5] Twitter Akshay [6] Twitter Owen [7] Twitter [8] [9] Twitter 1 1 Retweetability TweetLevel Twib Retweetist retweetradar retweetrank twittergrader

3 HITS A B C D E 3. Twitter Twitter 3. 1 Twitter ReTweet (ReTweet) (ReTweet) Twitter ( SA, TA) MA,1 MA, 2 MA, 3 ( SB, TB ) 2 MB,1 ( SC, TC ) MB, 2 MB, 3 ( S D, TD ) ( SE, TE ) MC,2 MC,1 MD,1 2 A M A,1, M A,2...M A,x (x A ) A (S A, T A ) (S A A T A A ) A M A,1 B (S A, T A ) M A,t = V + (Mi,j Ti) i F A (1) x k=1 S A = A,k x (2) T A = l l,t S l ) x (3) M i,j M A,t F A A M l,t A V (1) T i V (S A, T A ) (2) ME,1 ME,2

4 Algorithm 1 User-value 1: set all M A,t to V set all S i to 1 and pres i to set all T i to 1 pret i to 2: while S i pres i > ϵ and T i pret i > ϵ do 3: pres i S i V A B pret i T i 4: calculate M A,t by expression (1) calculate S i by expression (2) U calculate T i by expression (3) 5: end while 3 (3) S l S A T A Web HITS Algorithm ϵ A U U A U n(u) U U V V A B B A A U V n(v ) n(u) A n(v ) n(u) A n(v ) n(u) n(v ) n(u) U V U U V U n(v ) n(u) U A U A U n(v ) A n(u) V U A n(v )

5 Algorithm 2 User-value2 Require: Blogger set X 1: for all Blogger A in X do 2: set U A to 3: set n A (U) to 4: set n A (V ) to 5: for all B such that B is a follower of A do 6: U A U A {B} 7: n A (U) n A (U) + 1 8: for all C such that C is a follower of B and C / U do 9: n A (V ) n A (V ) + 1 1: end for 11: end for 12: calculate n A(V ) n A (U) 13: end for n(u)(a ) Algorithm2 1 2 (U ) U (V ) 3 n(v ) n(u) F F I (S, T ) I = (s S + (1 s) T ) F F ( < = s < = 1) s S T I (S, T ) F F 3. 2 I 4. 2 Twitter Twitter # #nhk #nhk NHK NHK 1163 ( ) P R W N C P = W N R = W C #nhk

6 人数 の数 精度 T 精度 S S T S (V ) T () V 1. 1 S V ( ) 1% 4 T ( ) 12% 45 S 4 T S S T T ( ) #nhk #tennouhai Twitter F F

7 精度 精度 s =.7 s =.5 s =.3 7 F F 9 I 精度 F F F F F F Twitter F F 2 8 F F F F 2 F F F F F F S T F F tirashiori hakkinton kuya KATOKICHIcoltd wakakit bottonbenjo magurohonsha burarimachi ESQ JPN bgyfromosaka yanagi moon gopochan tubyaki shiro asante8 UmiSola hashtagsjp kim take mac miyaby hajime13 mikeexpo I(s =.3) I(s =.5) I(s =.7) kuya kuya kuya hakkinton hakkinton UmiSola nkeisuke nkeisuke hajime13 bgyfromosaka UmiSola Mukunokiy miyaby hajime13 Otecchi 1 5 ID F F I s =.3,.5,.7 9 I S T F F F F I F F s S T 5 ID 1 1

8 I F F I F F 5. 2% % 4% % 3% % ( 2336, 214, , 2711) [1] K. Fujimura, T. Inoue, and M. Sugisaki. The eigenrumor algorithm for ranking blogs. In WWW Workshop on the Weblogging Ecosystem, 25. [2] A. Kritikopoulos, M. Sideri, and I. Varlamis. BlogRank: ranking weblogs based on connectivity and similarity features. In Proceedings of the 2nd international workshop on Advanced architectures and algorithms for internet delivery and applications, p. 8. ACM, 26. [3],,,,. blogger blog., Vol. 19, No. 2, pp , 27. [4] N. Agarwal, H. Liu, L. Tang, and P.S. Yu. Identifying the influential bloggers in a community. In Proceedings of the international conference on Web search and web data mining, pp ACM, 28. [5] Daniel M. Romero Bernardo A. Huberman and Fang Wu. Social networks that matter: Twitter under the microscope. First Monday, Vol. 14, No. 1-5, January 29. [6] 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 27 workshop on Web mining and social network analysis, pp ACM, 27. [7] Owen Phelan, Kevin McCarthy, and Barry Smyth. Using twitter to recommend real-time topical news. In RecSys, pp ACM, 29. [8],,.. The First Forum on Data Engineering and Information Management (DEIM), pp. A6 6, 29. [9],,,,.., Vol. 149, No. 18, pp. 2B 2, 29.