Anton Rubtsov

Vosstanovlenie Otsutsvuyushchikh Dannykh V Simvol'nykh Posledovatel'nostyakh: Vosstanovlenie Otsutstvushchikh Dannykh Anton Rubtsov Russian edition

V knige opisany metody vosstanovleniya otsutstvuyushchikh dannykh, kotorye rabotayut s simvol'nymi posledovatel'nostyami. Otsutstvuyushchaya chast' posledovatel'nosti vosstanavlivaetsya s ispol'zovaniem tol'ko toy informatsii, kotoraya soderzhitsya v samoy posledovatel'nosti (chastotnye slovari). Dlya vosstanovleniya dannykh ispol'zuyutsya sleduyushchie metody: kineticheskaya mashina Kirdina, matrichnoe predstavlenie chastotnykh slovarey i geneticheskie algoritmy. Obsuzhdeny dostoinstva i nedostatki etikh metodov. Vosstanovlenie otsutstvuyushchikh dannykh vypolnyaetsya bez ucheta semantiki. Ni odin metod vosstanovleniya otsutstvuyushchikh dannykh ne garantiruet vosstanovleniya tochnykh dannykh i ne daet otsenok tochnosti vosstanovlennykh dannykh. Mozhno govorit' tol'ko o tom, chto poluchennye dannye naibolee blizki k otsutstvuyushchim v silu togo ili inogo kriteriya. V nashem sluchae takim kriteriem yavlyaetsya uslovnaya entropiya. Uslovnaya entropiya opredelyaet kolichestvo ostayushcheysya entropii posle vosstanovleniya dannykh (ostayushcheysya neopredelennosti).