SERPWatcher SERPWatcher SERP Watcher SERP Watcher,

Size: px

Start display at page:

Download "SERPWatcher SERPWatcher SERP Watcher SERP Watcher,"

いっけいかいじ
4 years ago
Views:

1 SERPWatcher SERPWatcher SERP Watcher SERP Watcher, SERP Analysis of transition of ranking in SERP Watcher for social science research Fumiko NAKABE, Chiemi WATANABE, Naoko OYAMA, Kaoru TACHI, and Yoshifumi MASUNAGA Ochanomizu University Graduate School of Humanities and Sciences Otsuka, Bunkyou-ku, Tokyo, Japan Institute for Gender Studies, Ochanomizu University, Otsuka, Bunkyo-ku, Tokyo, Japan School of Social Informatics, Sagamihara-shi, Kanagawa, Japan nakabe@db.is.ocha.ac.jp, chiemi@is.ocha.ac.jp Abstract We develop the system named SERP Watcher for social science researchers to analyze transition of the ranking. The motivation of the system is based on our empirical knowledge that the ranking transitions of search engines seem to be related to social trends in the real world. SERP watcher collects ranking data of multiple search engines several times a week,and users can see the ranking transition of Web pages and they can also compare the rankings in multiple search engines. However,it is difficult that social scientists can find some relations between ranking transition and the social trends by seeing only rankings data. In this paper, we tried two types of analysis by using ranking transition data.one is extracting turning points of ranking transition by using interval least squares method,and the other is finding Web pages whose rank in a particular search engine is much difference from other search engines. Key words Web mining, Social science, Research methodology, Search engine, SERP 1. (search engine result page,serp) [1] [2] [3] SERP Watcher SERP Watcher SERP SERP SERPWatcher SERP SERPWatcher

2 2 3 SERP Watcher 4 6 SERPWatcher 7 2. Web [4] 2009 Web Web Web Web [5] Web [6] 2009 Web Web [7] 2005 Web Web 3. SERP Watcher SERPWatcher 3. 1 SERP Watcher SERP Watcher SERP Watcher 1 step1: step2: step3: SERPWatcher step1 1 SERP Watcher step SERP Watcher SERPWatcher SERP Watcher Google Yahoo!Japan msn infoseek baidu excite goo SERP 3 SERP Watcher Web SERPWatcher Web () Web Web 3 Web 4 Web Web 7

4. Web 3 Web SERPWatcher Web (1) (2) (3) 4 Web 4.

3 SERP Watcher SERPWatcher 2 30 20 5 Web 500 95 Web 300 250

3 4. Web 3 Web SERPWatcher Web (1) (2) (3) 4 Web Web Web SERPWatcher 2 Web SERP Watcher SERPWatcher Web Web (1) 1 score = 250 (1 ) (1) 2a rank rank score rank a a 3 a=0.1( ) a=0.3( ) a=0.1( ) Web

(2) lean lean = (ed score st score)/((ed

4 (2) lean lean = (ed score st score)/((ed wdate st wdate)/7)+1(2) a=0.03() [9] 2 Web () 6 google st date ed date st score ed score Web /1/1 2010/2/ Web

5 Web Web 8 ( ) Web 8 (B) Web Web 3 Web Web QandA Wikipedia ( 8 (D)) (C) Web 4 Web Web

5. 3 3 9 9 3 2009 Web 2009 [1],, : DEWS2006 17 4 ISSN 1347-4413 1B-i7 8p., 2006 3 [2], : Google Letters, Vol.5,No.2, pp.

6 Web 2009 [1],, : DEWS ISSN B-i7 8p., [2], : Google Letters, Vol.5,No.2, pp , [3] Naoko Oyama and Yoshifumi Masunaga: On the Trustworthiness and Trensparency of a Web Search Site examined using Gender-equal as a Search Keyword, Proceedings of APWeb2008, LNCS,Springer,April 2008 [4],, : DEIM 2009 A [5], : DEWS P-05, [6],, : Web Web DEIM 2009 B9-3, [7],, : DEWS C-i8 [8],,,,, : SERP Watcher DEIM 2009 D [9] Jon Kleinberg, Eva Tardos (,,, ) : pp [10] 6. SERP Watcher Web Web Web 20 (B)

Web [1] [2] [3] [4] [5] SupportVectorMachine SVM [6] [7] Google [11] Web

Web [1] [2] [3] [4] [5] SupportVectorMachine SVM [6] [7] Google [11] Web DEIM Forum 2009 E5-6 112-8610 2-1-1 112-8610 2-1-1 E-mail: {asami y,koba}@koba.is.ocha.ac.jp Web AJINOMOTO Easy Cooking Recipe Recommendation Considering User s Conditions Asami YAJIMA and Ichiro KOBAYASHI