DEIM Forum 2010 A3-3 Web Web 305 8550 1 2 305 8550 1 2 E-mail: s0813167@u.tsukuba.ac.jp, satoh@slis.tsukuba.ac.jp Web Web Web. Web Abstract Web-page Recommendation System based on the Keyword transitions through a Web Exploration Junya EDA and Tetsuji SATOH School of Library and Information Science,University of Tsukuba 1 2,Kasuga,Tsukuba,Ibaraki,305 0855 Japan Graduate School of Library Information and Media Studies,University of Tsukuba 1 2,Kasuga,Tsukuba,Ibaraki,305 0855 Japan E-mail: s0813167@u.tsukuba.ac.jp, satoh@slis.tsukuba.ac.jp In recent years, The number of web-pages are increasing considerably. User cannot find a desired page without much burdens. We propose an efficient method to recommend web pages, which user want to visit. We create the directed graph which node are keyword between pages and estimate keyword transition from connection strength between the keyword. We recommend the Web-pages that a user will read with the keyword which we estimated. We evaluated that we used the keyword which I extracted by the system. we could estimate the pattern of the Web reading action and confirmed that the keyword which I estimated by a keyword transition graph functioned effectively. Key words Web exploration logs, co-occurrence frequency, recommendation 1. Web. Web URL
Web Web Web URL Web 2 3 Web 4 5 6 2. Web 2. 1 Grineva [1] Wikipedia [2] Web [3] 2. 2 Web [4] web [5] [6] 2. 3 URL, 3. 3. 1 1 Web URL 3. 2 Web Simpson Web 4. 3 (directed acyclic graph)
3 2 Simpson(Apple, ipod) Simpson X Y X Y X Y X Y Simpson 1 Simpson(X, Y ) = X Y min( X, Y ) (1) 2 Simpson 3. 3 3. 3. 1 mecab 1 Termextruct [7] Simpson Appleipod 2 Simpson [8] Simpson (2) Simpson(X, Y ) X Y if X > k and Y > k, = min( X, Y ) 0 otherwise 4 (2) 1http://mecab.sourceforge.net/
6 5 3. 3. 2 Web 3 3 2 1 4 Simpson Simpson 3 2 iphone - iphone Simpson(,iphone) iphone 1 Apple Simpson(iphone,Aplle) iphone Applelog {Simpson(, iphone)} + log {Simpson(iphone, Apple)} 4 3 1 4. Web 5 Web URL Web 5 4. 1 6 Web URL URL Web 4. 2 Web HTML Web HTML Contentextruct 2 HTML HTML divtd Web 4. 3 Contentextruct Web Termextruct [7] Web 7 CNET Japan 3 MOBILE CHANNEL 2http://www.systemfriend.co.jp/node/326 3 CNET Japan http://japan.cnet.com/
7 Web 10 4 Google 4. 4 3. 3 Simpson Yahoo!WebsearchAPI 5 2 AND 2 Web,,, m n Simpson(m, n) Simpson 0.2 < = X < = 0.8 [8] Simpson 0.2 0.8 Google, simpson (2) k 3000 3000 4. 5 Web 8 3. 3. 2 Yahoo!WebserchAPI Web URL 4 CBS Interactive http://japan.cnet.com/mobile/story/0,3800078151,20406230,00.htm 5http://developer.yahoo.co.jp/ 8 Web 9 6 10 5. 5. 1 4. 3 Web Web 4 A. B. C. D. Web 5. 2 CNET Japan 1 3 CNET Japan 3. 1 5. 1 D. 6http://japan.cnet.com/news/media/story /0,2000056023,20394669,00.htm
9 2 B., 3 C..Web < > 3 5. 1 Web Web..Simpson < >,Simpson 3 6. 5. 3 5. 3. 1 1 3 () 4 1 D 2 B 1 2 3 Bing Microsoft Bing Cashback LAN Mbps NTT Microsoft Bing Hurt comscore 2 2 2 Bing Microsoft Google Wolfram Alpha 3 Android docomo PRO series NExus One docomo STYLE series NTT 3 C 4 A B C D A,B C,D 5. 3. 2
3 3 Forrester iphone iphone OS BlackBerry BlackBerry OS OS OS 2 iphone Google Google GS OS Andoroid 3 Nexus One Andoroid HTC Google 4 1. 1 2 3 U1 A B D U2 A B C U3 B A D U4 A B C U5 B C D U6 A B D 5 1 Bing 2 3 Google 8 5 Simpson Bing 2 3 Google 2 6 2 NTT 2 3 2 B () 9 B LANdocomoPROseries 7 3 2 3 iphone OS Google 6 2 NTT 3 D 2. 10 7 3 iphone OS Google 5. 4 4 Web Web 8 9, 10 Simpson 2 Web Bing iphoneos Web Web URL
8 1 A 2 2 3 6 6 8 5 Microsoft 5 Microsoft 8 Microsoft 5 Bing 4 Bing 4 Bing 4 4 2 2 4 2 1 1 9 2 B 2 2 3 7 docomoproseries 5 NTT 12 NTT 5 docomostyleseries 5 6 LAN 4 5 6 3 4 3 4 2 1 10 3 D 2 2 3 6 iphone 9 8 6 Google 7 Android 5 OS 4 OS 5 4 iphone 4 3 Nexus One 3 OS 3 2 iphone OS 1 2 URL 6. Web Simpson Web Simpson Web 4 Web Web [1] Maria Grineva, Maxim Grinev, and Dmitry Lizorkin. Extracting key terms from noisy and multi-theme documents. WWW2009, Vol. Mining for Semantics, pp. 661 670, 2009. [2],,,. Web., No. 2004-02, pp. 19 24, 2004. [3],,,. Web Web. DEWS2008, B2-3, 2008. [4],,. web., Vol. 17, No. 2, pp. 95 100, 2007. [5],.. JWEIN09, 2009. [6],,.., Vol. 49, No. 4, pp. 125 134, 2008. [7],,.., Vol. 10, No. 1, pp. 27 45, 2003. [8],,,,. Web., Vol. 20, No. 1, pp. 46 56, 2005.