Softversko inženjerstvo inteligentnih sistema
Software engineering of intelligent systems
Abstract
U ovoj disertaciji istražena je primena savremenih metoda softverskog inženjerstva u razvoju softvera otvorenog koda u oblasti inteligentnih sistema. Razvijen je opšti model softverskog frejmvorka i date su preporuke za vođenje projekata otvorenog koda u toj oblasti. Model frejmvorka i preporuke za vođenje projekata razvijeni su, i potvrđeni u praksi kroz razvoj dodatnih funkcionalnosti jednog izabranog softverskog frejmvorka za neuronske mreže. Izabrani softverki frejmvork za neuronske mreže je trenutno jedan od vodećih softvera u toj oblasti u svetu. Predložene metode razvoja softvera su opšteg karaktera, primenljive su i u drugim oblastima inteligentnih sistema, i odgovaraju na aktuelne probleme i specifičnosti razvoja softvera u tim oblastima. Problemi i specifičnosti se javljaju kao posledica toga što se često radi o softveru istraživačke namene koji treba da bude primenljiv i za praktične probleme, i kao posledica složenosti softvera koji istovremeno treba da bude jednostavan za ...korišćenje i modifikaciju. Predložene metode razvoja odnose se na specifikaciju zahteva, planiranje razvoja, projektovanje, implementaciju i održavanje, i date su u vidu preporuka koje su se pokazale kao dobra praksa. Pored ovog teorijskog i metodološkog doprinosa, kroz disertaciju su razvijena i konkretna unapređenja izabranog frejmvorka iz oblasti inteligentnih sistema, koja značajno unapređuju njegovu praktičnu primenu...
This dissertation survey the application of modern software engineering methods for open source software development in the field of intelligent systems. Main results are a general model of software framework and recommendations for managing open source projects in that field. Framework model and recommendations for the project management were developed and validated in practice through the development of additional features of a selected software framework for neural networks. Selected neural network framework is one of the leading Java open source neural network frameworks. Suggested software development methods are also applicable in other fields of intelligent systems, and they are response to problems and specific requirements in this field. Most problems are due the fact that it is often software for research purposes, that should also be applicable for practical problems, and complexity of the software that also needs to be easy to use and modify. Proposed development methods ar...e related to requirements specification, planning, design, implementation and maintenance. They are given in the form of recommendations that have been proven as a good practice. In addition to these theoretical and methodological contributions, this dissertation also provides practical contribution in the form of new features for one selected framework in the field of intelligent systems, which greatly enhance its practical application...
Keywords:
softverski frejmvork / softver otvorenog koda / razvoj softvera / inteligentni sistemi / software framework / software development / open source software / intelligent systemsSource:
2012Publisher:
- Univerzitet u Beogradu, Fakultet organizacionih nauka
Funding / projects:
- Infrastructure for Technology Enhanced Learning in Serbia (RS-MESTD-Integrated and Interdisciplinary Research (IIR or III)-47003)
URI
http://eteze.bg.ac.rs/application/showtheses?thesesId=598https://nardus.mpn.gov.rs/handle/123456789/3051
https://fedorabg.bg.ac.rs/fedora/get/o:6760/bdef:Content/download
http://vbs.rs/scripts/cobiss?command=DISPLAY&base=70036&RID=515131034
https://rfos.fon.bg.ac.rs/handle/123456789/62
Collections
Institution/Community
Fakultet organizacionih naukaTY - THES AU - Ševarac, Zoran PY - 2012 UR - http://eteze.bg.ac.rs/application/showtheses?thesesId=598 UR - https://nardus.mpn.gov.rs/handle/123456789/3051 UR - https://fedorabg.bg.ac.rs/fedora/get/o:6760/bdef:Content/download UR - http://vbs.rs/scripts/cobiss?command=DISPLAY&base=70036&RID=515131034 UR - https://rfos.fon.bg.ac.rs/handle/123456789/62 AB - U ovoj disertaciji istražena je primena savremenih metoda softverskog inženjerstva u razvoju softvera otvorenog koda u oblasti inteligentnih sistema. Razvijen je opšti model softverskog frejmvorka i date su preporuke za vođenje projekata otvorenog koda u toj oblasti. Model frejmvorka i preporuke za vođenje projekata razvijeni su, i potvrđeni u praksi kroz razvoj dodatnih funkcionalnosti jednog izabranog softverskog frejmvorka za neuronske mreže. Izabrani softverki frejmvork za neuronske mreže je trenutno jedan od vodećih softvera u toj oblasti u svetu. Predložene metode razvoja softvera su opšteg karaktera, primenljive su i u drugim oblastima inteligentnih sistema, i odgovaraju na aktuelne probleme i specifičnosti razvoja softvera u tim oblastima. Problemi i specifičnosti se javljaju kao posledica toga što se često radi o softveru istraživačke namene koji treba da bude primenljiv i za praktične probleme, i kao posledica složenosti softvera koji istovremeno treba da bude jednostavan za korišćenje i modifikaciju. Predložene metode razvoja odnose se na specifikaciju zahteva, planiranje razvoja, projektovanje, implementaciju i održavanje, i date su u vidu preporuka koje su se pokazale kao dobra praksa. Pored ovog teorijskog i metodološkog doprinosa, kroz disertaciju su razvijena i konkretna unapređenja izabranog frejmvorka iz oblasti inteligentnih sistema, koja značajno unapređuju njegovu praktičnu primenu... AB - This dissertation survey the application of modern software engineering methods for open source software development in the field of intelligent systems. Main results are a general model of software framework and recommendations for managing open source projects in that field. Framework model and recommendations for the project management were developed and validated in practice through the development of additional features of a selected software framework for neural networks. Selected neural network framework is one of the leading Java open source neural network frameworks. Suggested software development methods are also applicable in other fields of intelligent systems, and they are response to problems and specific requirements in this field. Most problems are due the fact that it is often software for research purposes, that should also be applicable for practical problems, and complexity of the software that also needs to be easy to use and modify. Proposed development methods are related to requirements specification, planning, design, implementation and maintenance. They are given in the form of recommendations that have been proven as a good practice. In addition to these theoretical and methodological contributions, this dissertation also provides practical contribution in the form of new features for one selected framework in the field of intelligent systems, which greatly enhance its practical application... PB - Univerzitet u Beogradu, Fakultet organizacionih nauka T1 - Softversko inženjerstvo inteligentnih sistema T1 - Software engineering of intelligent systems UR - https://hdl.handle.net/21.15107/rcub_nardus_3051 UR - t-3542 ER -
@phdthesis{ author = "Ševarac, Zoran", year = "2012", abstract = "U ovoj disertaciji istražena je primena savremenih metoda softverskog inženjerstva u razvoju softvera otvorenog koda u oblasti inteligentnih sistema. Razvijen je opšti model softverskog frejmvorka i date su preporuke za vođenje projekata otvorenog koda u toj oblasti. Model frejmvorka i preporuke za vođenje projekata razvijeni su, i potvrđeni u praksi kroz razvoj dodatnih funkcionalnosti jednog izabranog softverskog frejmvorka za neuronske mreže. Izabrani softverki frejmvork za neuronske mreže je trenutno jedan od vodećih softvera u toj oblasti u svetu. Predložene metode razvoja softvera su opšteg karaktera, primenljive su i u drugim oblastima inteligentnih sistema, i odgovaraju na aktuelne probleme i specifičnosti razvoja softvera u tim oblastima. Problemi i specifičnosti se javljaju kao posledica toga što se često radi o softveru istraživačke namene koji treba da bude primenljiv i za praktične probleme, i kao posledica složenosti softvera koji istovremeno treba da bude jednostavan za korišćenje i modifikaciju. Predložene metode razvoja odnose se na specifikaciju zahteva, planiranje razvoja, projektovanje, implementaciju i održavanje, i date su u vidu preporuka koje su se pokazale kao dobra praksa. Pored ovog teorijskog i metodološkog doprinosa, kroz disertaciju su razvijena i konkretna unapređenja izabranog frejmvorka iz oblasti inteligentnih sistema, koja značajno unapređuju njegovu praktičnu primenu..., This dissertation survey the application of modern software engineering methods for open source software development in the field of intelligent systems. Main results are a general model of software framework and recommendations for managing open source projects in that field. Framework model and recommendations for the project management were developed and validated in practice through the development of additional features of a selected software framework for neural networks. Selected neural network framework is one of the leading Java open source neural network frameworks. Suggested software development methods are also applicable in other fields of intelligent systems, and they are response to problems and specific requirements in this field. Most problems are due the fact that it is often software for research purposes, that should also be applicable for practical problems, and complexity of the software that also needs to be easy to use and modify. Proposed development methods are related to requirements specification, planning, design, implementation and maintenance. They are given in the form of recommendations that have been proven as a good practice. In addition to these theoretical and methodological contributions, this dissertation also provides practical contribution in the form of new features for one selected framework in the field of intelligent systems, which greatly enhance its practical application...", publisher = "Univerzitet u Beogradu, Fakultet organizacionih nauka", title = "Softversko inženjerstvo inteligentnih sistema, Software engineering of intelligent systems", url = "https://hdl.handle.net/21.15107/rcub_nardus_3051, t-3542" }
Ševarac, Z.. (2012). Softversko inženjerstvo inteligentnih sistema. Univerzitet u Beogradu, Fakultet organizacionih nauka.. https://hdl.handle.net/21.15107/rcub_nardus_3051
Ševarac Z. Softversko inženjerstvo inteligentnih sistema. 2012;. https://hdl.handle.net/21.15107/rcub_nardus_3051 .
Ševarac, Zoran, "Softversko inženjerstvo inteligentnih sistema" (2012), https://hdl.handle.net/21.15107/rcub_nardus_3051 .