{"id":12285,"date":"2023-05-28T14:10:57","date_gmt":"2023-05-28T11:10:57","guid":{"rendered":"https:\/\/bestessayhomework.com\/tr\/?p=12285"},"modified":"2023-05-28T14:10:57","modified_gmt":"2023-05-28T11:10:57","slug":"ozyinelemeli-arama-modeli","status":"publish","type":"post","link":"https:\/\/bestessayhomework.com\/tr\/ozyinelemeli-arama-modeli\/","title":{"rendered":"\u00d6zyinelemeli Arama Modeli\u00a0"},"content":{"rendered":"

At\u0131f Da\u011f\u0131l\u0131mlar\u0131n\u0131n Ampirik Karakterizasyonu<\/span><\/strong><\/h3>\n

A\u015fa\u011f\u0131da, al\u0131nt\u0131 da\u011f\u0131l\u0131mlar\u0131n\u0131 parametrele\u015ftirmek i\u00e7in ayr\u0131k g\u00fc\u00e7 yasas\u0131n\u0131 kullan\u0131yoruz. Kar\u015f\u0131l\u0131k gelen kaymay\u0131 ve e\u011fimi yakalayan w ve \u03bd uydurma parametrelerinin zamana ba\u011fl\u0131l\u0131\u011f\u0131n\u0131 g\u00f6sterir. W kaymas\u0131 zamanla artar ve \u223c10 y\u0131l sonra doyuma ula\u015f\u0131r. \u03bd \u00fcss\u00fc s\u00fcrekli olarak azal\u0131r ancak doyuma ula\u015fmaz. Bu, 25 y\u0131l sonra bile at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n dura\u011fan olmad\u0131\u011f\u0131 ve kuyruklar\u0131n\u0131n geli\u015fmeye devam etti\u011fi anlam\u0131na gelir.<\/span><\/p>\n

B\u00f6ylece, at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n uzun zaman s\u0131n\u0131r\u0131nda s\u0131n\u0131rlay\u0131c\u0131 bir \u015fekil ald\u0131\u011f\u0131na dair yayg\u0131n inan\u0131\u015f\u0131n aksine, at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n dura\u011fan olmad\u0131\u011f\u0131n\u0131 ve 25 y\u0131l sonra bile s\u0131n\u0131rlay\u0131c\u0131 bir \u015fekle ula\u015fmad\u0131\u011f\u0131n\u0131 g\u00f6rd\u00fck. Dura\u011fan olmama nedeni ise y\u00fcksek at\u0131f alan makalelerdir. A\u015fa\u011f\u0131da g\u00f6sterdi\u011fimiz gibi, bu ka\u011f\u0131tlar sonunda ka\u00e7ak haline gelir.<\/span><\/p>\n

Asl\u0131nda, erken at\u0131f da\u011f\u0131l\u0131mlar\u0131, log-normal veya \u03bd = 4-5 ile ayr\u0131k \u00fcstel kanun da\u011f\u0131l\u0131m\u0131 ile uydurulabilir. B\u00fcy\u00fck \u00fcs, bu da\u011f\u0131l\u0131mlar\u0131n “geleneksel” oldu\u011funu, yani kuyruklar\u0131n\u0131n yaln\u0131zca k\u00fc\u00e7\u00fck bir rol oynad\u0131\u011f\u0131n\u0131 g\u00f6sterir. da\u011f\u0131l\u0131m\u0131n ortalamas\u0131n\u0131 ve varyans\u0131n\u0131 belirlemede.<\/span><\/p>\n

Zaman ge\u00e7tik\u00e7e ve makaleler daha fazla al\u0131nt\u0131 toplad\u0131k\u00e7a, al\u0131nt\u0131 da\u011f\u0131l\u0131mlar\u0131 sa\u011fa kayar. S\u0131radan gazetelerin at\u0131f kariyeri sona erdi\u011finde \u223c10 y\u0131l sonra bu de\u011fi\u015fim sona erer. Daha sonra, at\u0131flar \u00e7o\u011funlukla da\u011f\u0131l\u0131m\u0131n kuyru\u011funu olu\u015fturan y\u00fcksek at\u0131fl\u0131 makaleler taraf\u0131ndan toplan\u0131r.<\/span><\/p>\n

Da\u011f\u0131l\u0131m\u0131n e\u011fimi daha kademeli hale gelir, \u00e7\u00fcnk\u00fc zaman ge\u00e7tik\u00e7e kuyruk sa\u011fa do\u011fru h\u0131zla hareket ederken da\u011f\u0131l\u0131m\u0131n geri kalan\u0131 yava\u015flar ve sonunda hareketsiz kal\u0131r. Kuvvet yasas\u0131 \u00fcss\u00fc \u03bd’nin zamanla azalmas\u0131n\u0131n nedeni budur.<\/span><\/p>\n

Sonunda, da\u011f\u0131t\u0131m\u0131n kuyru\u011fu yaln\u0131zca at\u0131f kariyeri s\u00fcresiz olarak uzayan ka\u00e7ak makalelerden olu\u015fur. Kuyruk sa\u011fa do\u011fru hareket etmeye devam ederken, da\u011f\u0131l\u0131m\u0131n geri kalan\u0131 hi\u00e7bir zaman doygunlu\u011fa gelmeyecek ve sonunda i\u00e7b\u00fckey olacak \u015fekilde sabit kal\u0131r.<\/span><\/p>\n

Peki, i\u00e7b\u00fckey kuyru\u011fu olan al\u0131nt\u0131 da\u011f\u0131l\u0131mlar\u0131 neden bu kadar nadir? At\u0131f da\u011f\u0131l\u0131m\u0131n\u0131n i\u00e7b\u00fckey bir kuyru\u011funu g\u00f6zlemlemek i\u00e7in \u00e7ok uzun bir zaman penceresi ve \u00e7ok say\u0131da ka\u00e7ak makale i\u00e7eren \u00e7ok b\u00fcy\u00fck bir veri seti gerektirdi\u011fini iddia ediyoruz.<\/span><\/p>\n

Sunulan t\u00fcm setler aras\u0131nda, 40.195 Fizik makalesinden olu\u015fan en b\u00fcy\u00fck set bile bu ama\u00e7 i\u00e7in yetersizdir. Bununla birlikte, 1980-1989’da yay\u0131nlanan 10 kat daha b\u00fcy\u00fck fizik makalesinin, g\u00f6sterildi\u011fi gibi i\u00e7b\u00fckey kuyru\u011fu ortaya \u00e7\u0131kar\u0131r.<\/span><\/p>\n

\u00d6zyinelemeli Arama Modeli\u00a0<\/span><\/strong><\/h3>\n

\u00d6nceki yaz\u0131lar\u0131m\u0131zda g\u00f6sterdi\u011fimiz gibi, modelimiz \u00e7e\u015fitli disiplinler i\u00e7in at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131 ba\u015far\u0131l\u0131 bir \u015fekilde yeniden \u00fcretir. Burada, \u00f6l\u00e7\u00fcmler i\u00e7in eri\u015filemeyen zaman penceresine ekstrapolasyon yapmak i\u00e7in bu modeli kullan\u0131yoruz.<\/span><\/p>\n

Say\u0131sal olarak sim\u00fcle edilmi\u015f at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n dura\u011fan hale gelmedi\u011fini ve yay\u0131nland\u0131ktan 50 y\u0131l sonra bile geli\u015fmeye devam etti\u011fini g\u00f6sterir. Bu, ampirik parametrenin, yani at\u0131f da\u011f\u0131l\u0131m\u0131n\u0131n s\u00f6zde \u00fcst kural kuyru\u011funun \u00fcss\u00fcn\u00fcn zamanla neden azald\u0131\u011f\u0131n\u0131 ve doyuma ula\u015fmad\u0131\u011f\u0131n\u0131 a\u00e7\u0131klar.<\/span><\/p>\n

At\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n genel \u015feklinden modelin hangi parametreleri sorumludur? Erken at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n uygunluk da\u011f\u0131l\u0131m\u0131n\u0131 taklit etti\u011fini zaten fark etmi\u015ftik. Fizik makaleleri i\u00e7in, birincisi \u03bc = \u22121.48, \u03c3 = 1.12 ile lognormal olarak veya \u03bd = 4 ile Waring da\u011f\u0131l\u0131m\u0131 olarak temsil edilebilir. Her neyse, her iki uyum da d\u0131\u015fb\u00fckey da\u011f\u0131l\u0131mla sonu\u00e7lan\u0131r.<\/span><\/p>\n

Say\u0131sal sim\u00fclasyon, zaman ge\u00e7tik\u00e7e at\u0131f da\u011f\u0131l\u0131m\u0131n\u0131n sa\u011fa kayd\u0131\u011f\u0131n\u0131 ve kuyru\u011funun d\u00fczle\u015fti\u011fini, \u00f6yle ki 2 < \u03bd < 3 ile bir kuvvet yasas\u0131 ba\u011f\u0131ml\u0131l\u0131\u011f\u0131 gibi g\u00f6r\u00fcnd\u00fc\u011f\u00fcn\u00fc g\u00f6sterir. Bu kuvvet kanunu benzeri ba\u011f\u0131ml\u0131l\u0131k 3 i\u00e7in ge\u00e7erlidir. \u2013Yay\u0131nlanmas\u0131ndan 20 y\u0131l sonra ve bu nedenle al\u0131nt\u0131 da\u011f\u0131l\u0131mlar\u0131 ayr\u0131 yetki yasas\u0131na ba\u015far\u0131yla uyarlanm\u0131\u015ft\u0131r.<\/span><\/p>\n

\n

Recursive fonksiyon \u00f6rnekleri<\/span>
\n\u00d6zyineleme<\/a> nedir<\/span>
\nC# recursive fonksiyon \u00d6rnekleri<\/span>
\nJava recursive fonksiyon<\/span>
\nRecursive fonksiyon C<\/span>
\n\u00d6zyinelemeli algoritma nedir<\/span>
\nAlgoritma Analizi yerine koyma metodu<\/span>
\n\u00d6zyinelemeli fonksiyonlar<\/span><\/p>\n

\n

Ancak, daha uzun zaman aral\u0131klar\u0131 ve b\u00fcy\u00fck ka\u011f\u0131t k\u00fcmeleri i\u00e7in durum farkl\u0131d\u0131r. Sim\u00fclasyonun t = 50 y\u0131l olarak ekstrapolasyonu, uzun zaman s\u0131n\u0131r\u0131nda da\u011f\u0131l\u0131m\u0131n d\u00fcz kalmak yerine i\u00e7b\u00fckey hale geldi\u011fini g\u00f6sterir. Bu, ara zamanlarda at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n d\u00fcz kuyru\u011funun yaln\u0131zca ge\u00e7ici bir \u015fekil oldu\u011fu anlam\u0131na gelir.<\/span><\/p>\n

Dura\u011fan olmayan at\u0131f da\u011f\u0131l\u0131mlar\u0131ndan modelimizin hangi \u00f6zelli\u011fi sorumludur? A\u015fa\u011f\u0131da, bunun, at\u0131f say\u0131s\u0131 ba\u011f\u0131ml\u0131l\u0131\u011f\u0131na g\u00f6re artan dolayl\u0131 al\u0131nt\u0131 olas\u0131l\u0131\u011f\u0131 P0 oldu\u011funu g\u00f6steriyoruz. Ger\u00e7ekte, ger\u00e7ek P0(K) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131n\u0131 bir sabitle, P0 = 0.54 ile de\u011fi\u015ftirdi\u011fimiz say\u0131sal sim\u00fclasyon ger\u00e7ekle\u015ftirdik.\u00a0<\/span><\/p>\n

Modelimiz hi\u00e7bir parametrenin K’ye ba\u011fl\u0131 olmad\u0131\u011f\u0131 do\u011frusal bir modele indirgenir. Do\u011frusal modelin erken at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131 olduk\u00e7a iyi a\u00e7\u0131klarken, ge\u00e7 da\u011f\u0131t\u0131mlarda sefil bir \u015fekilde ba\u015far\u0131s\u0131z oldu\u011funu g\u00f6sterir. Do\u011frusal model, t\u00fcm zamanlar i\u00e7in ayn\u0131 e\u011fime sahip at\u0131f da\u011f\u0131l\u0131mlar\u0131 verirken, \u00f6l\u00e7\u00fclen at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n e\u011fimi zamanla s\u00fcrekli olarak azal\u0131r.<\/span><\/p>\n

At\u0131f \u00d6mr\u00fc<\/span><\/strong><\/h3>\n

G\u00fc\u00e7 yasas\u0131 al\u0131nt\u0131 da\u011f\u0131l\u0131mlar\u0131n\u0131 iddia eden \u00f6nceki ampirik \u00e7al\u0131\u015fmalar, bu da\u011f\u0131l\u0131mlar\u0131n \u00f6l\u00e7ekten ba\u011f\u0131ms\u0131z oldu\u011funu ima etti. Elbette at\u0131flar kesikli ve negatif olmad\u0131\u011f\u0131ndan, bu da\u011f\u0131l\u0131mlar\u0131n ortalama at\u0131f say\u0131s\u0131 gibi do\u011fal bir \u00f6l\u00e7e\u011fi vard\u0131r.<\/span><\/p>\n

“\u00d6l\u00e7eksiz at\u0131f da\u011f\u0131l\u0131m\u0131” ile genellikle M taraf\u0131ndan belirlenen mikroskobik \u00f6l\u00e7e\u011fin yan\u0131 s\u0131ra makroskobik \u00f6l\u00e7e\u011fin olmamas\u0131 kastedilir. \u00d6zellikle, 1984’te yay\u0131nlanan Fizik makaleleri i\u00e7in, 2008 y\u0131l\u0131na kadar birikmi\u015f ortalama al\u0131nt\u0131 say\u0131s\u0131 M = 26’d\u0131r. \u00f6l\u00e7ek, \u015eekil 8.1c’de g\u00f6r\u00fclebilir ve K < M i\u00e7in da\u011f\u0131l\u0131m\u0131n d\u0131\u015fb\u00fckey k\u0131sm\u0131ndan K > M i\u00e7in d\u00fcz kuyru\u011fa ge\u00e7i\u015fe kar\u015f\u0131l\u0131k gelir.<\/span><\/p>\n

Bu d\u00fcz kuyruk K\/M \u223c 3’ten K\/M = 230’a kadar uzan\u0131r. Bu iki say\u0131 aras\u0131ndaki b\u00fcy\u00fck fark \u00f6l\u00e7eksiz da\u011f\u0131l\u0131m\u0131n bir g\u00f6stergesi olarak kabul edilir. Ancak bu sadece bir g\u00f6stergedir, bir kan\u0131t de\u011fildir. A\u015fa\u011f\u0131da, at\u0131f da\u011f\u0131l\u0131mlar\u0131n\u0131n, g\u00f6zden ka\u00e7an makalelerin ba\u015flang\u0131c\u0131yla ili\u015fkili bir Kr \u00f6l\u00e7e\u011fine sahip oldu\u011funu g\u00f6steriyoruz. Bu gizli \u00f6l\u00e7ek, at\u0131f da\u011f\u0131l\u0131mlar\u0131nda neredeyse hi\u00e7 g\u00f6r\u00fclmez, ancak makalelerin at\u0131f y\u00f6r\u00fcngelerini analiz etti\u011fimizde a\u00e7\u0131k\u00e7a ortaya \u00e7\u0131kar.<\/span><\/p>\n

Bu ama\u00e7la, 1984’te yay\u0131nlanan 40.195 Fizik makalesi i\u00e7in say\u0131sal sim\u00fclasyonumuza geri d\u00f6n\u00fcyoruz ve tek tek makalelerin at\u0131f yollar\u0131na odaklan\u0131yoruz. Bu y\u00f6r\u00fcngeleri analiz ederek, \u00f6l\u00e7\u00fclen al\u0131nt\u0131 y\u00f6r\u00fcngeleri i\u00e7in yapt\u0131\u011f\u0131m\u0131z gibi al\u0131nt\u0131 \u00f6mr\u00fcn\u00fc \u03c40 bulduk. Kar\u015f\u0131l\u0131k gelen eskime oran\u0131n\u0131 g\u00f6sterir, \u0393sim = 1\/\u03c40. Sim\u00fcle edilmi\u015f \u0393sim, \u0393meas ile olduk\u00e7a uyumludur.<\/span><\/p>\n

\u00d6zellikle, modelimiz belirli Kr’de azalan \u0393 (K) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131n\u0131 ve at\u0131f \u00f6mr\u00fcn\u00fcn sapmas\u0131n\u0131 olduk\u00e7a iyi yeniden \u00fcretir. Bu Kr, ka\u00e7aklar\u0131n ba\u015flang\u0131c\u0131n\u0131 karakterize eder ve \u00f6l\u00e7\u00fclen al\u0131nt\u0131 da\u011f\u0131l\u0131m\u0131nda d\u00fcz kuyru\u011fun i\u00e7b\u00fckey kuyru\u011fa d\u00f6n\u00fc\u015ft\u00fc\u011f\u00fc nokta olarak ay\u0131rt edilebilir.<\/span><\/p>\n

Bu \u015fa\u015f\u0131rt\u0131c\u0131 \u0393 (K) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131, P0(K) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131ndan kaynaklanmaktad\u0131r. Asl\u0131nda, P0 = const ile do\u011frusal bir model kullanan say\u0131sal sim\u00fclasyon, \u0393sim’in K ile yaln\u0131zca k\u00fc\u00e7\u00fck K < 20’de azald\u0131\u011f\u0131n\u0131 g\u00f6sterir. Bu \u0393sim(K) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131, al\u0131nt\u0131lar\u0131n ayr\u0131k olmas\u0131 ger\u00e7e\u011finden kaynaklan\u0131r.<\/span><\/p>\n

Asl\u0131nda, bir makalenin at\u0131f oran\u0131n\u0131n son at\u0131f oran\u0131na ba\u011fl\u0131 oldu\u011fu ayr\u0131 bir Hawkes s\u00fcrecini tan\u0131mlar. Bu ba\u011f\u0131ml\u0131l\u0131k, dalgalanmalar\u0131 art\u0131ran pozitif bir geri besleme sa\u011flar ve k\u00fc\u00e7\u00fck K i\u00e7in \u0393sim(K) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131n\u0131 azaltmaktan sorumludur.<\/span><\/p>\n

Ancak, K > 20 i\u00e7in bu tamamen istatistiksel etki ortadan kalkar ve \u0393s i m (K ) ba\u011f\u0131ml\u0131l\u0131\u011f\u0131 bir d\u00fczl\u00fc\u011fe ula\u015f\u0131r. Bu, t\u00fcm K i\u00e7in artan K ile d\u00fc\u015fmeye devam eden \u00f6l\u00e7\u00fclen \u0393meas ile \u00e7eli\u015fir. Bu nedenle, do\u011frusal model K < 20 i\u00e7in \u00f6l\u00e7\u00fcmlerimizi a\u00e7\u0131klarken, K > 20 i\u00e7in onlarla uyu\u015fmaz.\u00a0<\/span><\/p>\n

\n\n
<\/div>\n\n<\/div>\n \n