ãã®è¨äºã¯ abicky.net ã® NMF: Non-negative Matrix Factorizationï¼éè² å¤è¡åå ååè§£ï¼ ã«ç§»è¡ãã¾ãã
id:naoya ããã®Latent Semantic Indexing ã®è¨äºã«è§¦çºããã¦ããã1é±éã»ã©ã¡ããã¡ããè¦ã¦ããè¡åã®è¿ä¼¼è¨ç®ææ³ã«ã¤ãã¦æ¸ãã¦ã¿ããããã§ããããã®ã¯åèª-ææ¸è¡å(ã©ã®åèªãã©ã®ææ¸ã«åºã¦ãããã®å ±èµ·è¡å)ãè³¼å ¥è -ã¢ã¤ãã è¡å(ã©ã®äººãã©ã®æ¬ãè²·ã£ããã¨ããæ¨è¦ã¨ã³ã¸ã³ã§ä½¿ãè¡å)ããã¼ã¸-ãªã³ã¯è¡å(ã©ã®ãã¼ã¸ããã©ã®ãã¼ã¸ã«ãªã³ã¯ãåºã¦ãããããããã¯ãªã³ã¯ãããã£ã¦ããããPageRank ãªã©ãã¼ã¸ã®ã©ã³ãã³ã°ã®è¨ç®ã«ä½¿ã)ãã¨ãã£ããããªè¡åãè¨ç®ããã¨ãã大è¦æ¨¡è¡åã ã¨è¨ç®éã»è¨æ¶ã¹ãã¼ã¹ã¨ãã«è¨å¤§ãªã®ã§ãäºåã«ããç¨åº¦è¨ç®ãã¦ãããã®ã§ããã°ãã§ããã ãå°ãããã¦ãããã(ããã¦å¯è½ãªãã°ç²¾åº¦ãä¸ããã)ãã¨ããææ³ã§ããã è¡åã®å§ç¸®ã«ã¯å ã®è¡åã A (mè¡nå)ã¨ãã㨠A = USV^T ã¨ããããã«3ã¤ã«å解ãããã¨ãå¤ãããã
ã¦ã§ãã°ã©ãã®ãªã³ã¯è§£æã«ãããã¼ã¸ã®è©ä¾¡ã¨è¨ãã° PageRank ãèåã§ãããããã²ã¨ã¤ Jon Kleinberg ã«ãã HITS (Hyperlink-induced topic search)ãæåã§ããæåã®è«æ Authoritative Sources in a Hyperlinked Environment 㯠1999å¹´ã§ããIIR ã® 21ç« ã§ããã® PageRank 㨠HITS ã«ã¤ãã¦ã®è§£èª¬ãããã¾ããã HITS HITS ã¯ã¦ã§ããã¼ã¸ã®è©ä¾¡ã«äºã¤ã®è»¸ãç¨ãã¾ããä¸ã¤ã authority ã¹ã³ã¢ãããä¸ã¤ã hub ã¹ã³ã¢ã§ãã ä¾ãã°ãPerl ã®æ å ±ã欲ãããã¨ããæ¤ç´¢è¦æ±ã«å¯¾ãã¦ã¯ CPAN ã éçºè ã§ãã Larry Wall ã®ãã¼ã ãã¼ã¸ãªã©ãéè¦åº¦ã®é«ããã¼ã¸ãã¨æãã¾ãããããã®ãã¼ã¸ã¯ãPerl ã«é¢ãã¦ä¿¡é ¼ã§ããæ å ±æºãã¨ãããã¨
æ å ±æ¤ç´¢ã«ããããã¯ãã«ç©ºéã¢ãã«ã§ã¯ãææ¸ããã¯ãã«ã¨ã¿ãªãã¦ç·å½¢ç©ºéã§ãããæ±ãã¾ãããã®ææ¸ãã¯ãã«ã¯ãææ¸ã«å«ã¾ããåèªã®åºç¾é »åº¦ãªã©ãæåã«åãã¾ããçµæã以ä¸ã®ãããªåèªææ¸è¡å (term document matrix) ãå¾ããã¾ãã d1 d2 d3 d4 Apple 3 0 0 0 Linux 0 1 0 1 MacOSX 2 0 0 0 Perl 0 1 0 0 Ruby 0 1 0 3 ãã®åèªææ¸è¡åã«å¯¾ãã¦å ç©ã«ããé¡ä¼¼åº¦ãªã©ã®è¨ç®ãè¡ã£ã¦ãæ å ±è¦æ±ã«é©åããææ¸ãæ¢ãã®ããã¯ãã«ç©ºéã¢ãã«ã«ããæ¤ç´¢ã¢ãã«ã§ãã è¦ã¦ã®éããåèªææ¸è¡åã®æ¬¡å æ°ã¯ç´¢å¼èªã®ç·æ°ã§ããææ¸ãå¢ããã°å¢ããã»ã©æ¬¡å ã¯å¢å ããå¾åã«ããã¾ããä¾ãã°ç´¢å¼èªã100ä¸èªãã£ã¦æ¤ç´¢å¯¾è±¡ã®ææ¸ã 1,000ä¸ä»¶ããã¨ã100ä¸æ¬¡å * 1,000ä¸ã¨ãã大ããã®è¡åãæ±ããã¨ã«ãªãã¾ãããå
ã¡ãã£ã¨é£ã°ãã¦ï¼å ã«IIR18ç« ãèªãã§ã¿ãï¼åèªææ¸è¡åãç¹ç°å¤å解ãã¦æ°ãã空éã§ãã¯ãã«ç©ºéã¢ãã«ã使ãã¨ããLSIã®è©±ï¼ ãã¼ã¸æ°ãå°ãªãã£ãã®ã§ï¼éãå·®ãã¦ç¿»è¨³ããã¦ã¿ãï¼ããã«æ°å¼ãå¤ãã®ã§TeXã§æ¸ãã¦ã¿ãï¼ããã¾ã§æ¥ãããã ãããã¨ABåã®æªãçãåºã¦ï¼æ°å¼ãæ¼ç¿ãå ¨é¨è¨³ãã¦ã¿ãï¼ã¤ãã«ãã¨ãªã£ã¦ãã£ã¦ãã¾ã£ãï¼ä»ã¯å ¬éãã¦ããï¼ã§ãåçã¯ãã¦ããªãï¼ã¾ã ãã£ã¤ãã®é¨åãããã®ã§ãã¤ãã¤ã¨ãã¼ã¸ã§ã³ã¢ãããã¦ãã¾ãï¼ Introduction to information retrieval: 18 Matrix decomposition and latent semantic indexingï¼åè¨³ï¼ å¤§ä½1ãã¼ã¸1æéï¼ãã¤ãã¤å¤ãªã¹ããã¦3æ¥éããããããã¾ããï¼å¦ãå¿ã§ãç²¾èªããã®ã§ï¼ã¨ã¦ãç解ãæ·±ã¾ãã¾ããï¼ãã£ããèªãã®ã翻訳ä½æ¥ãã¨ã¦ã楽ããã£ãã®ã§ï¼ãªã
ç¾æç¹ã§ã¯ãããã°ãã¼ã¿ã§ã¯ãªãããµã³ããªã³ã°ããããã¼ã¿ãæ±ã£ã¦ããã®ã§ãããã大è¦æ¨¡ããã¼ã¿ã§ãããã¨ã«ã¯éããªããããããå¦ä½ã«é«éã«å¦çãããã«é ãæ©ã¾ãã¦ãã¾ãã ãã®æ¬ã¯ããã®æ©ã¿ã解決ããå©ãã«ãªãã¾ããã å®éã«ã¯ãããããã®åæç¥èãå¿ è¦ã§ããã対象ã¨ããåé¡ãéçºç°å¢ãèæ ®ããªããã°ãªãã¾ããã§ããã é¨åçãªèª¿æ´ããã§ãã¦ãã¾ããããããã§ããå½åã¯è´å½çã¨ãæãããå®è¡é度ããä½ã¨ãç¾å®çãªé度ã¾ã§æã¦ãããã¨ãã§ãã¾ããã 以ä¸ã«ããã®æ¬ãèªãéã«åèã«ãããã¼ã¸ãå種å®è£ é¢é£ã®ãã¼ã¸ãªã©ã¸ã®ãªã³ã¯ãï¼åå¿é²ãå ¼ãã¦ï¼è¨ãã¦ããã¾ãã ãªãããåèã«ããããã¼ã¸ã§ãã®ã§ãå¿ ãããæ¬æ¸ã«ã¦è§£èª¬ããã¦ããé ç®ã¨ã¯éãã¾ããã â»å®è£ ã³ã¼ãã®é ç®ã«è¨è¼ããã©ã¤ã»ã³ã¹ã¯èª¿æ»æã®ãã®ã§ãããæªç¢ºèªã»æªè¨è¼ã®ãã®ãããã¾ããå©ç¨ã«éãã¦ã¯ããããã確èªãå¿ è¦ã§ãã â å ¨è¬
æ¬æ¥ï¼PFI ã»ããã¼ã«ã¦ãä¹±æãã¼ã¿æ§é ã®ææ°äºæ âMinHash 㨠HyperLogLog ã®æè¿ã®é²æ©âãã¨ããã¿ã¤ãã«ã§è©±ãããã¦ãããã¾ããï¼ã¹ã©ã¤ãã¯ä»¥ä¸ã«ãªãã¾ãï¼ Ustream ã®é²ç»ãããã¾ãï¼ http://www.ustream.tv/recorded/48151077 å 容ã¨ãã¦ã¯ï¼ä»¥ä¸ã®æä½ãå¹ççã«è¡ãããã®éåã«é¢ãããã¼ã¿æ§é (Sketch) ã®æè¿ã®é²æ©ãç´¹ä»ãã¾ããï¼ éåã®é¡ä¼¼åº¦ã®æ¨å® (Jaccard ä¿æ°) éåç°ãªãæ°ã®æ¨å® (distinct counting) ã©ã¡ããéè¦ãã¤åºç¤çãªæä½ã§ï¼b-bit MinHash ã HyperLogLog ãªã©ï¼æ¢ã«å®ç¨çãªææ³ãææ¡ããã¦ããï¼å®éã«ã使ããã¦ãã¾ãï¼ãããï¼2014 å¹´ã«ãªã£ã¦ï¼Odd Sketch ã HIP Estimator ã¨ããï¼ããããããã«æ¹åããææ³ãç«ã¦ç¶
転置ã¤ã³ããã¯ã¹ï¼ã¦ãã¡ã¤ã³ããã¯ã¹ãInverted indexï¼ã¨ã¯ãå ¨ææ¤ç´¢ãè¡ã対象ã¨ãªãææ¸ç¾¤ããåèªã®ä½ç½®æ å ±ãæ ¼ç´ããããã®ç´¢å¼æ§é ãããã転置索å¼ã転置ãã¡ã¤ã«ãéå¼ãç´¢å¼ãªã©ã¨ãå¼ã°ããã æ å ±å¦çãã¯ããã¸ã«ããã転置ã¤ã³ããã¯ã¹ã¨ã¯ãåèªãæ°åã¨ãã£ãå 容ããããããå«ã¾ãã¦ãããã¼ã¿ãã¼ã¹ãããã¥ã¡ã³ã群ã¸ã®ãããã³ã°ãä¿æããã¨ãããã¤ã³ããã¯ã¹åãã¼ã¿æ§é ã§ãããããã¥ã¡ã³ã群ã¸ã®ãããã³ã°ã®å ´åãæ¤ç´¢ã¨ã³ã¸ã³ãå®ç¾ãããã転置ã¤ã³ããã¯ã¹ãã¡ã¤ã«ã¯ãã¤ã³ããã¯ã¹ã¨ããããã¯ãã¼ã¿ãã¼ã¹ã¨å¼ãã ã»ãããµããããå ´åããããã¾ããæ¤ç´¢ãã¼ãåèªï¼æååï¼ã§ãããé£æ³é åã®å¤ãä½ç½®æ å ±ã§ããå ´åãããã·ã¥ãã¼ãã«ã®å½¢æ ãåããã¨ãããã 転置ã¤ã³ããã¯ã¹ã«ã¯å¤§ããåãã¦2éãã®ææ³ããããã¬ã³ã¼ãåä½è»¢ç½®ã¤ã³ããã¯ã¹ï¼record level inverted in
第6åãå ¨ææ¤ç´¢ãå®è£ ããã½ã¼ã¹ã³ã¼ããèªãã å貫 義人 æ¾æ ç« å¼ TISæ ªå¼ä¼ç¤¾ SonicGarden 2009/9/3 åªããããã°ã©ãã¯ã³ã¼ããæ¸ãã®ã¨åãããããã³ã¼ããèªã¿ããªããªãã¦ã¯ãªããªããåªããã³ã¼ããèªããã¨ã§ãèªèº«ã®ã¹ãã«ãä¸éããã®ã ï¼ç·¨éé¨ï¼ ãããããªã¼ãã³ã½ã¼ã¹ã®ç¤¾å SNSãSKIPãã使ã£ãã³ã¼ããªã¼ãã£ã³ã°ãæçµåã¨ãªãã¾ãããRailsã®åºæ¬çãªæ§æããããã¹ãã³ã¼ããRSpecã®æ¸ãæ¹ã¨ãã£ãå 容ã«å ããååã¯OpenIDãRailsã§æ´»ç¨ããå¿ç¨ç·¨ã¾ã§ãã³ã¼ãã¨ã¨ãã«å¦ãã§ãã¾ããã æçµåã¨ãªãä»åã¯ãSKIPã®ç®çæ©è½ã®1ã¤ã§ããå ¨ææ¤ç´¢ãæ±ãã¾ããæçµåã«ãµãããããå 容ãé«åº¦ãªãã®ã«ãªã£ã¦ãã¾ãããããã¾ã§ãã¤ãããããã ããèªè ã®çãã¾ã§ããã°ãååã«ç解ã§ããå 容ã ã¨æãã¾ãã SKIPã«ãããå ¨ææ¤ç´¢æ©è½ã§ã¯ãä»»æã®æ¤ç´¢ãã¼ã¯ã¼ã
ã¯ããã« æ¤ç´¢ã¨ã³ã¸ã³ã¨èãã¨ãã¿ãªããã¯ä½ãæãæµ®ãã¹ãã§ããããï¼ GoogleãYahoo!ãªã©ã®æ¤ç´¢ãã¼ã¸ãæãæµ®ãã¹ãæ¹ãã»ã¨ãã©ã ã¨æãã¾ããè¿å¹´ã¯ããããä¼æ¥ã®åªåã«ãã£ã¦æ¤ç´¢ã¨ã³ã¸ã³ã¨ãããã®ãé常ã«èº«è¿ã«ãªããç§ãã¡ã®çæ´»ã«æ¬ ãããªããã®ã¨ãªãã¤ã¤ããã¾ãã ããããæ¤ç´¢ã¨ã³ã¸ã³ã¨ä¸è¨ã§è¨ã£ã¦ããä¸è¨ã®ãããªä¸è¬ã®æ¹ã ã¸ã®UIï¼ã¦ã¼ã¶ã¤ã³ã¿ã¼ãã§ã¼ã¹ï¼ãæãå ´åãããã°ããã®UIã®è£å´ï¼ããã¯ã¨ã³ãï¼ã«ããã·ã¹ãã ãæãå ´åãããã¾ãã æ¬é£è¼ã§ã¯ãGoogle,Yahoo!ãªã©ã代表ã¨ããæ¤ç´¢ã¨ã³ã¸ã³ã®è£å´ã®ããã¿ã«çç®ããæ¤ç´¢ã¨ã³ã¸ã³ã¨ããã·ã¹ãã ã®ã¢ã¼ããã¯ãã£ããã®å é¨ã§ä½¿ããã¦ãããã¼ã¿æ§é ãã¢ã«ã´ãªãºã ããè¿å¹´ã®ææ³ãç 究äºä¾ãªã©ã交ãã¦è§£èª¬ãã¦ããããã¨æã£ã¦ãã¾ãã æ¤ç´¢ã¨ã³ã¸ã³ã¨ã¯ æ¤ç´¢ã¨ã³ã¸ã³ã«ã¯ããã¾ãã¾ãªç¨®é¡ãããã¾ããGoogleã®Webæ¤ç´¢ã®ã
A van Emde Boas tree (Dutch pronunciation: [vÉn ËÉmdÉ ËboËÉs]), also known as a vEB tree or van Emde Boas priority queue, is a tree data structure which implements an associative array with m-bit integer keys. It was invented by a team led by Dutch computer scientist Peter van Emde Boas in 1975.[1] It performs all operations in O(log m) time (assuming that an bit operation can be performed in cons
ãé«éæåå解æã®ä¸çãã¨ãã大å¤ãã°ãããæ¬ãçºå£²ããããããã¨æ·å± ãé«ãæ¬ã§ã¯ããã®ã§èªãåã«ç¥ã£ã¦ããã¨ãããããªãã¨ãæ¸ãã¦ããã ãé«éæåå解æãã¨ã¯ æ¬æ¸ã§ããé«éæåå解æã¨ããã®ã¯ä¸»ã«2ã¤ã®ãã¨ãæãã¦ãããã²ã¨ã¤ã¯ãã¼ã¿ãå§ç¸®ãã¦å°ãããã¦ãã£ã¹ã¯ããã¡ã¢ãªãã¡ã¢ãªãããã£ãã·ã¥ã¨ããããã«ããé«éãªè¨æ¶è£ ç½®ã§æ±ãã¾ããããã¨ãã話ãããã²ã¨ã¤ã¯ãã¼ã¿æ§é ã工夫ãããã¨ã§è¤éãªæä½ãããããé«éã«æ±ãã¾ãããã¨ãã話ãã¤ã¾ããå§ç¸®ãã®è©±ã¨ãå¹ççãªãã¼ã¿æ§é ãã®è©±ãããã¨èãã¦ããã¨è¯ãã ãã¼ã¯ã¼ãã¯3㤠ãªãã«ãæ¸ãã¦ããããã©ãæ¬æ¸ã主ã«æ±ãã®ã¯ãBWTããç°¡æ½ãã¼ã¿æ§é ããã¦ã§ã¼ãã¬ããæ¨ãã®3ã¤ãå ·ä½çã«ã¯ãBWTãããå§ç¸®ãã«é¢ãã£ã¦ãã¦ãã¦ã§ã¼ãã¬ããæ¨ãããå¹ççãªãã¼ã¿æ§é ãã«é¢ãã£ã¦ããããç°¡æ½ãã¼ã¿æ§é ãã¯åºæ¬çãªéå ·ã¨ãã¦æ¬æ¸ã®è²ã ãªã¨ããã§åºã¦
ãªãªã¼ã¹ãé害æ å ±ãªã©ã®ãµã¼ãã¹ã®ãç¥ãã
ææ°ã®äººæ°ã¨ã³ããªã¼ã®é ä¿¡
å¦çãå®è¡ä¸ã§ã
j次ã®ããã¯ãã¼ã¯
kåã®ããã¯ãã¼ã¯
lãã¨ã§èªã
eã³ã¡ã³ãä¸è¦§ãéã
oãã¼ã¸ãéã
{{#tags}}- {{label}}
{{/tags}}