Modelovanje i implementacija sistema za podršku vrednovanju publikovanih naučno-istraživačkih rezultata

Abstract

Cilj – Prvi cilj istraživanja je kreiranje modela podataka i implementacija informacionog sistema zasnovanog na modelu za potrebe vrednovanja publikovanih naučno-istraživačkih rezultata. Model bi bio primenjen u CRIS UNS informacionom sistemu, kao podrška sistemu vrednovanja. Drugi cilj istraživanja je utvrđivanje u kojoj meri i na koji način se može automatizovati proces evaluacije koji se zasniva na različitim pravilima i pravilnicima. Metodologija – Kako bi se definisalo proširenje CERIF modela neophodno je bilo identifikovati različite aspekte podataka koji su prisutni u evaluaciji naučno-istraživačkih publikacija. Stoga, zarad potreba istraživanja, odabrana su i analizirana su dokumenta koja predstavljaju različite nacionalne pravilnike, okvire i smernice za evaluaciju. Za modelovanje specifikacije arhitekture sistema za vrednovanje korišćeni su CASE alati koji su bazirani na objektno-orijentisanoj metodologiji (UML 2.0). Za implementaciju proširenja CERIF modela u CRIS UNS sistemu korišćena je Java platforma i tehnologije koji olakšavaju kreiranje veb aplikacija kao što su AJAX, RichFaces, JSF itd. Pored navedene opšte metodologije za razvoj softverskih sistema korišćeni su primeri dobre prakse u razvoju informacionih sistema. To se pre svega odnosi na principe korišćene u razvoju institucionalnih repozitorijuma, bibliotečkih informacionih sistema, informacionih sistema naučno-istraživačke delatnosti, CRIS sistema, sistema koji omogućuju evaluaciju podataka itd. Ekspertski sistem koji bi podržao automatizaciju procesa evaluacije po različitim pravilnicima odabran je na osnovu analize postojećih rešenja za sisteme bazirane na pravilima i pregleda naučne literature. Rezultati – Analizom nacionalnih pravilnika i smernica dobijen je skup podataka na osnovu kojeg je moguće evaluirati publikovane rezultate po odabranim pravilnicima. Razvijen je model podataka kojim se predstavljaju svi podaci koji učestvuju u procesu evaluacije i koji je kompatibilan sa CERIF modelom podataka. Predloženi model je moguće implementirati u CERIF kompatibilnim CRIS sistemima, što je potvrđeno implementacijom informacionog sistema za vrednovanje publikovanih naučno-istraživačkih rezultata u okviru CRIS UNS. Ekspertski sistem baziran na pravilima može biti iskorišćen za potrebe automatizacije procesa evaluacije, što je potvrđeno predstavom i implementacijom SRB pravilnika u Jess sistemu baziranom na pravilima. Praktična primena –Zaključci proizašli iz analize pravilnika (npr. poređenje sistema i definisanje metapodataka za vrednovanje) se mogu primeniti pri definisanju modela podataka za CERIF sisteme i za sisteme koji nisu CERIF orijentisani. Sistem za podršku vrednovanju publikovanih naučno-istraživačkih rezultata je implementiran kao deo CRIS UNS sistema koji se koristi na Univerzitetu u Novom Sadu čime je obezbeđeno vrednovanje publikovanih naučno-istraživačkih rezultata za različite potrebe (npr. promocije u naučna i istraživačka zvanja, dodele nagrada i materijalnih sredstava, finansiranje projekata, itd.), po različitim pravilnicima i komisijama. Vrednost – Dati su metapodaci na osnovu kojih se vrši vrednovanje publikovanih rezultat istraživanja po raznim nacionalnim pravilnicima i smernicama. Dat je model podataka i proširenje CERIF modela podataka kojim se podržava vrednovanje rezultata istraživanja u CRIS sistemima. Posebna prednost pomenutih modela je nezavisnost istih od implementacije sistema za vrednovanje rezultata istraživanja. Primena predloženog proširenje CERIF modela u CRIS sistemima praktično je pokazana u CRIS sistemu Univerziteta u Novom Sadu. Sistem za vrednovanje koji se bazira na proširenju CERIF modela pruža i potencijalnu interoperabilnost sa sistemima koji CERIF model podržavaju. Implementacijom informacionog sistema za vrednovanje, vrednovanje naučnih publikacija je postalo olakšano i transparentnije. Potvrda koncepata da se ekspertski sistemi bazirani na pravilima mogu koristiti za automatizaciju vrednovanja, otvara totalno novi okvir za implementaciju informacionih sistema za podršku vrednovanja postignutih rezultata istraživanja.

Aim – The first aim of the research was creation of data model and implementation of information system based on the proposed model for the purpose of evaluation of published research outputs. The model is applied in CRIS information system to support the system for evaluation. The second objective was determination of the manner and extent in which the evaluation process that is based on different rules and different rulebooks could be automated. Methodology - In order to define the extension of the CERIF model, it was necessary to identify the various aspects of data which is relevant in evaluation of scientific research publications. Therefore, documents representing different national regulations, frameworks and guidelines for evaluations were selected and analyzed. For the modeling of the system architecture, CASE tools were used, which are based on object-oriented methodology (UML 2.0). To implement the extension of the CERIF model within the CRIS UNS system, JAVA platform and technologies that facilitate creation of web applications such as AXAJ and RichFaces were used. In addition to this general methodology for development of software systems, best practice examples from the information systems development are also used. This primary refers to the principles used in development of institutional repositories, library information systems, information systems of the scientific-research domain, CRIS systems, systems that enable evaluation of data, etc. The expert system that supports automation of the evaluation process by different rulebooks was selected based on analysis of the existing solutions for rule based systems and examination of scientific literature. Results - By analysis of the national rulebooks and guidelines, a pool of data was gathered, which served as a basis for evaluation of published results by any analyzed rulebook. A data model was developed, by which all data involved in the evaluation process can be represented. The proposed model is CERIF compatible. The proposed model can be implemented in CERIF compatible CRIS systems, which was confirmed by the implementation of an information system for evaluation of published scientific research results in CRIS UNS. An expert system based on rules can be used for the needs of automation of the evaluation process, which was confirmed by the presentation and implementation of the Serbian Rulebook by Jess. Practical application - The conclusions raised from the analysis of rulebooks (e.g. Comparison of systems and defining metadata for evaluation) can be applied in defining the data model for CERIF systems and for systems that are not CERIF oriented. The system for support of evaluation of published scientific research results was implemented as part of the CRIS UNS system used at the University of Novi Sad, thus providing evaluation of published scientific research results for different purposes (e.g. promotion in scientific and research titles, assignment of awards and material resources, financing of projects, etc.), according to different rulebooks and commissions. Value – Metadata is provided on which basis the evaluation of published research results by various national rulebooks and guidelines is conducted. A data model and an expansion of the CERIF data model that supports the evaluation of the research results within CRIS systems are given. A special advantage of these models is their independence of the implementation of the system for evaluation of research results. The application of the proposed extension of the CERIF model into CRIS systems practically is demonstrated in the CRIS system of the University of Novi Sad. The system that implements an expansion of the CERIF model provides a potential interoperability with systems that support CERIF model. After the implementation of the information system for evaluation, the evaluation of scientific publications becomes easier and more transparent. A confirmation of the concept that the expert systems based on rules can be used in automation of the evaluation process opens a whole new framework for implementation of information systems for evaluation.