20.500.12556/DKUM-36499 UPORABA NOSQL PODATKOVNIH BAZ ZA GENERIRANJE POROČIL DETEKTORJA PLAGIATOV USAGE OF NOSQL DATABASES FOR PLAGIAT DETECTOR REPORT GENERATION V diplomskem delu smo predstavili koncept podatkovnih baz NOSQL. Našteli smo osnovne skupine, jih opisali in našteli njihove pripadnike. Podrobneje smo se posvetili opisu podatkovnega sistema Apache Cassandra, njegovim lastnostim, delovanju in arhitekturi. Posebej so nas zanimale splošne lastnosti Cassandre, kot so elastičnost, razpoložljivost in ujemljivost podatkov ter zanesljivost in zmogljivost sistema v realnem okolju. V praktičnem delu diplomskega dela smo razvili sistem, ki uporablja Appache Cassandra za generiranje poročil o vsebinski podobnosti med dokumenti. Iskanje plagiatov je časovno zahteven proces, saj z naraščanjem števila dokumentov narašča tudi časovna zahtevnost iskanja. Cassandrina elastičnost in podatkovni model sta idealna rešitev za takšno iskanje. Za izvedbo preizkusa smo uporabili obstoječo bazo izvlečkov (angl. hash), ki smo jih pridobili iz dokumentov Digitalne knjižnice Univerze v Mariboru. To smo prenesli v Cassandrino gručo desetih strežnikov. Čase generiranja poročil smo primerjali s časi, ki smo jih dobili pri generiranju poročil iz podatkovne baze MS SQL. Pokazali smo, da Appache Cassandra 2,2-krat hitreje generira poročila in je neobčutljiva za izpade strežnikov. In this diploma thesis we have introduced the concept of NOSQL databases. We described some of the NOSQL groups and represented their members. Next chapters are dedicated to the detailed description of Apache Cassandra, its attributes, architecture and how it works in general. We were especially interested in general attributes of Cassandra, like elastic scaling, availability and data consistency, reliability and performance with a real work load. Further, we developed a system which uses Apache Cassandra to generate reports on content similarity of documents. The search of plagiarism is a time consuming process and for each document added, time complexity grows. The elasticity and the data model of Apache Cassandra are an ideal solution for this problem. For the realization of our performance test, we used an existing database of hash values from the Digital library of the University of Maribor. We transferred those values to our cluster of ten servers and compared the measured report generation times from Apache Cassandra cluster to the measured report generation times from the MS SQL database. We have shown, that Apache Cassandra is generating reports 2,2 times faster and that it is insensitive for server failures. podatkovne baze NOSQL Appache Casandra teorem CAP detekcija plagiatov NSQL databases Appache Casandra CAP theorem plagiat detection true true false [M. Dietner] Slovenski jezik Angleški jezik Diplomsko delo 2012-06-11 12:09:18 2012-06-15 11:25:21 2022-04-20 08:47:14 2024-01-02 03:07:10 2012 0 Maribor 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 004.65(043.2) 16235030 URN:SI:UM:DK:C09O9IT5 UNI_Dietner_Mario_2012.pdf UNI_Dietner_Mario_2012.pdf 1 63694AFB7A00C4FEDE821DA3402CE43F 8f0c02bdbc461a69fc7c362ed6604aad844e8ecb835ab717123e4759c55d4618 10567683-7c08-11eb-bb7a-00155d0001ca 20.500.12556/dkum/db51f7f6-101a-4740-a262-bb457f7637f3 https://dk.um.si/Dokument.php?lang=slv&id=47629 Fakulteta za elektrotehniko, računalništvo in informatiko 0 0 0