5

Size: px

Start display at page:

Download "5"

关肖浦
5 years ago
Views:

1 Query Processor 理 Dataset Preparor DataBase CiteSeer 論 CiteSeer 料 Crawler PreProcessor DataBase 裡 Data Preparor Ranker 論裡力論 Website Locator Query Processor Dataset Preparor Crawler Ranker PreProcessor Website Locator Presentation Module Clustering Module DataBase

2 Presentation Module 1. Query Processor 理路來 CGI 例 " " "%2B" " " "%3D" 參數 Dataset Preparor 數 2. Dataset Preparor 論 Crawler CiteSeer 論 DataBase 裡不 CiteSeer 來 PreProcessor 理 DataBase Data Preparor 4.1 Dataset Collection 裡 Base Set Finding 了兩 (1) CiteSeer top papers (2) Clustering-based Base Set Finding Base Set Extension 了 (1) Extend cite papers (2) Extend cited papers (3) Extend cite and cited papers (4) Feedback papers from important conference and journal base set Extension 數不更利 Clustering 來 base set Clustering Module 來若 Dataset Crawler CiteSeer 料料都 3. Ranker 論裡力論 50

3 論 4.3 Paper-based Ranking Algorithms 力力略略裡了力 (1) Basic Strategy PageRank TimedPageRank HITS (2) Asynchronous Ranking PageRank TimedPageRank HITS (3) Synchronous Ranking PageRank TimedPageRank HITS 參數更不參數更更力 Dataset Feedback Dataset Preparor Website Locator 4. Website Locator Ranker 來力 Crawler Google 利 4.5 Presentation Module 5. Presentation Module: 力 Paper-based Ranking Algorithm PageRank TimedPageRank 列力論 HITS 列 Authority 論 Hub 論力論了論論力力 51

4 Website Locator 6. Clustering Module: 論度裡了 (1) Cite information only (2) Cited information only (3) Cite and cited information 來論行裡來 base set hierarchical clustering 論論 Dataset Preparor base set 7. Crawler 路料都不不不來料裡料 CiteSeer 利 CiteSeer 理來來料 PreProcessor Google Google 理理料 PreProcessor ISI ISI 料裡料理 PreProcessor DBLP 兩 DBLP 料裡料例料 DBLP 理 PreProcessor 8. PreProcessor: 理 Crawler 料來料 HTML 料不 Database 52

5 Input: q: user query sw: way of getting base set ew: way of extending base set tn: number of extension Output: S: proper paper set Data_prepare{ } S := {}; S getbaseset(q, sw); Do{ S extendbaseset(ew, S); }until tn times output S; 5.2 Dataset preporor algorithm 料 Database 裡 CiteSeer 來論利 4.2 論識來 9. DataBase: 料裡 CiteSeer 論了 base set 論 CiteSeer 論 CiteSeer 不 request request 都路來路塞 CiteSeer request 都理度了料都 Database 裡料不 CiteSeer request Database 裡來

6 Input: q: user query, sw: way of getting base set Output: B: base set getbaseset{ } B := {}; B crawler(q); // crawler return CiteSeer top t papers If( sw == 1 ){ // CiteSeer top t papers Return B; }elsif( sw == 2 ){ // Clustering-based base set finding B clustering(b); } Return B; 5.3 Base set finding algorithm Perl C FreeBSD 料 Mysql 論裡路 Query Processor 利 Perl 參數 Dataset Preparor 理 1. CGI 來料 2. 參數來 3. HTML 4. 料 Dataset Preparor DataSet Preparor Query Processor 來參數論 CiteSeer clustering Base set 利論

7 Input: B: base set, ew: way of extending base set Output: S: proper paper set extendbaseset{ } S := {}; if( ew == 1 ){ // extend cite papers S P where p P, s S, and s cite p ; }elsif( ew == 2 ){ // extend cited papers S P where p P, s S, and p cite s ; }elsif( ew == 3 ){ // extend cite and cited papers S P where p P, s S, and s cite p or p cite s ; }elsif( ew == 4 ) // extend from important conference or journal S P where p P, s S, and s cite p or p cite s ; } Ranker Set feedback flag; Return S ; 5.4 Base set extension algorithm Ranker 裡了 Webpage Ranking Algorithm 說 1. PageRank (4.1) d TimedPageRank (4.5) (4.6) d 0.85 w Aging() [47] 裡 Aging() 論 1 論 0.5 了落 0.5 NOW Time( pi ) Aging( pi ) = 1 0.5* (5.1) PERIOD PERIOD 裡 20 說年 20 年 0.5 NOW 年 Time(p i ) 論 p i 55

8 論力論路論 HITS 論裡 Math::Cephes::Matrix 量論力 Math::Cephes::Matrix Website Locator 4.5 Google 列列料裡路 5.1 列了度 2 說 Google 連度 2 兩不列類連列 Publications Book Courses About Papers Profile Address Research Service Teaching Research interests Slide Tutorials Awards Fax Phone Contact information Projects Journal Conference Professional Activities Personal Current Activities Office Personal Information Vitae Research group Bio Research areas Education Academic Info 56

9 5.2.5 Clustering Module Clustering 1. 不力論論數 5 論數 5 clustering 論裡 2. 力論 Key Paper 論論論數論數 1 Key Paper 3. Soft Cluster Key Paper 論 Key Paper 度 soft cluster 裡度量 co-citation coupling co-citation 數 0 coupling 數 0 兩論 soft cluster 4. Hierarchical clustering soft clusters 兩 soft clusters 度 A B 兩 soft clusters 度 A B A + B A B (5.2) A B 論度 Clustering 來 base set clusters 數量 base set 數 5. cluster 裡力論 cluster 論 cluster 裡論數數論 base set Crawler 裡 crawler 料都 Crawler 路來路料兩 57

10 5.5 ISI 列 5.6 ISI 列論 1. 兩連來連兩來論 (1) 連連兩例利例 href="../../indices/a-tree/f/fang:hai.html" 例 href="../../indices/a-tree/f/fang:hai.html" uni-trier.de /~ley/db/conf/aaai/aaai2002.html 裡 trier.de/~ley/db/indices/a-tree/f/fang:hai.html 58

5.7 DBLP 論料 (2) 路都利 CGI 來 CGI 兩 GET 例 data mining http://citeseer.ist.psu.edu/cis?

11 5.7 DBLP 論料 (2) 路都利 CGI 來 CGI 兩 GET 例 data mining POST 數來 5.5 料利來立連數來 5.6 GET 參數裡 POST 數都來數來 2. 路料 HTML 裡不料不連不不料料來 5.7 論都 " " 論論論 " " 59

12 5.8 DBLP 論料 "," HTML 裡不 5.8 <li> " " 了 "," 連論 <li> 論論論 " " "," 連連裡來連利裡 Crawler CiteSeer Google DBLP ISI 料 Database 裡說 60

13 5.9 CiteSeer 1. CiteSeer Crawler CiteSeer 行 (1) CiteSeer CiteSeer 兩 documents Citations documents 論 citations CiteSeer CGI GET data mining &cs=1 裡 5.9 CiteSeer Citations 數 Expected citations 數 Hub HITS 裡 Hub Usage 數 Date 索 61

14 (b) (a) 5.10 CiteSeer 論 (1) (a) (b) 5.11 CiteSeer 論 (2) 62

15 裡 citations 來 base set (2) 5.9 論料論更論論來論 (3) 論論論 5.11(a) 論 5.11(a) 論 5.11(a) 論 5.11(a) 論 PDF 5.10(a) 論 5.11(b) 論 5.10(b) 論都來 CiteSeer 料論 PDF 連來連來 CiteSeer 裡不論都 PDF 5.10 不 5.10 PS.gz PS PDF IMAGE 論 PDF PS.gz PS IMAGE 論裡連 CiteSeer 裡論論 5.12 類來 (4) 論 5.10(b) 論 5.13 裡論料 Database 裡 63

16 5.12 CiteSeer 論 5.13 CiteSeer 論 64

2. Google Google CiteSeer 都 CGI GET Google http://www.google.com/search?hl=en&ie=utf-8&q=data+mining &btng=google+search 5.14 連料 Google 連 <a href=http://www.kdnuggets.

17 2. Google Google CiteSeer 都 CGI GET Google &btng=google+search 5.14 連料 Google 連 <a href= onmousedown="return clk(this,'res',2)"> 連裡了了連 Google Google 來率 3. DBLP DBLP 行 (1) 列 DBLP 列索 26 裡 a Google 65

18 5.15 DBLP (1) index-a.html index-b.html 索 b 了 DBLP 裡 (2) 裡列了 5.16 連料 (3) 論料 5.7 (4) 列 DBLP 立列 5.17 料都來 Database 裡 (5) 料 DBLP 列 DBLP CGI POST 數裡 DBLP 兩 66

19 5.16 DBLP (2) 列 5.18 列 5.18 都列料料 4. ISI ISI 料 ISI 10 料 " " 10 料 5.5 " " 利 CGI POST ISI 滑來 x.next y.next 兩錄數 old_page 來數料 67

20 5.17 DBLP 列 5.18 列 68

21 5.2.7 PreProcessor CiteSeer 料論兩 1. 論 CiteSeer 論串理例 {IEEE} {T}rans. on {E}volutionary {C}omputation Proc. of 1995 Int'l Conf. on Very Large Data Bases ({VLDB}'95), Z{\ Data \& Knowledge Engineering {SPIRE}/{CRIWG} Fortran Journal VLDB'97 Proc. IEEE Int. Conf. Robot. Autom. (ICRA) 例 CiteSeer 料亂料行理 4.2 論裡 (1) 例 {IEEE} {T}rans. on {E}volutionary {C}omputation IEEE Transactions on Evolutionary Computation Data \& Knowledge Engineering Data and Knowledge Engineering (2) 論 4.2 類論了例 "Proc. of 1995 Int'l Conf. on Very Large Data Bases ({VLDB}'95), Z{\" "International Conference on Very Large Data Bases" 2. CiteSeer 來串串例 69

22 "Mohamed G. Elfeky and Amani A. Saad and Souheir A. Fouad" 利 "and" 來 Database 裡力論都論料料裡立了料 PaperDB 論欄 pid checkname title sid time abstract keyword pdf memo 5.2 abstract keywords SourceDB 論欄 sid checkname name abb-name pub-place issn site last-site period kind memo 5.3 AuthorDB 料欄 aid checkname name site dept domain memo 5.4 dept domain 5.2 料 PaperDB PaperDB pid 論 id 來不論 checkname 論 Database 來論 title 論 sid 論 sourcedb 裡 id time 論 abstract 論 ( ) keywords 論 ( ) pdf 論 PDF memo 70

23 5.3 料 SourceDB SourceDB sid 論 id 來不論 checkname 論 Database 來論 name 論 abb-name 論 pub-place issn site 論 last-site 論 period 論兩 kind 類 memo 5.4 料 AuthorDB AuthorDB aid id 來不 checkname Database 來 name site dept ( ) domain 領 ( ) memo 了料論論利料來 A2P 論了論欄 aid pid ord 5.5 A2S 論論欄 aid sid 來 A2P PaperDB 參了 71

24 立來料料 A2P aid pid ord A2P AuthorDB aid PaperDB pid 論例 5.6 料 A2S aid sid AuthorDB aid SourceDB sid A2S 5.7 料 Citation cite cited Citation 論論 pid 論 pid Citation 論論欄 cite cited 兩 PaperDB pid 5.7 論料來 CiteSeer CiteSeer 論料裡論料 Database 裡都論料利 CiteSeer 論不立了 linkdb linkdb CiteSeer 裡論連 PaperDB 裡論欄 link pid 路 Dataset 5.8 Database ER-model

25 5.8 料 linkdb link pid CiteSeer PaperDB 裡 pid linkdb Free Software Packages 裡利路不 1. LWP[37] LWP Library libwww-perl-5 Gisle Aas Roy Fielding Perl 4 libwww library 來良 Web Client Programming Perl 來路料裡 crawler crawler 料例 request header 裡 "Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; KKman2.0)" request KKman 來理度 CGI GET POST 行數 request 流便利 LWP 路料 2. Math::Cephes::Matrix[38] 數了 73

26 checkname abb-name sid name pub-place last-site issn SourceDB site period memo kind publish pid checkname title Citation PaperDB time sid write abstract pdf memo keywords memo domain dept AuthorDB site name aid checkname 5.19 料 ER-model 量 74

27 來量 HITS 3. Bit::Vector[2]: C Boolean Vector Vector bit Boolean Vector 了列兩 Boolean Vector Boolean Vector 數例 8 10 Boolean Vector 例 shift rotate Perl 裡列列裡不裡什 32byte Perl 留若 5000x5000 列 2Gigabyte 裡列 0 1 連若 Perl 列浪 Bit::Vector 來省利 Boolean Vector 列列 75

untitled

untitled 不料金類料不料利零利料不利料不料量料流來料料行都兩類來例不例 58 兩不類不冷狀不不惡料不行冷行料利論力不力行不不裡例車車量金都不易易不易易 59 類易廉度量料料量度省例例類料車零 60 類輪拉來狀料都裂來煉不粒粒狀