Please use this identifier to cite or link to this item:
https://rfos.fon.bg.ac.rs/handle/123456789/63| Title: | Ekspertni sistemi i sistemi za izveštavanje Expert systems and reporting systems |
Authors: | Tomić, Bojan | Contributors: | Devedžić, Vladan Starčević, Dušan Đurić, Dragan Jovanović, Jelena Obradović, Ivan |
Keywords: | poslovna pravila;poslovna inteligencija;mehanizam za objašnjavanje;ključni indikatori poslovanja;izveštavanje;ekspertni sistem;reporting;key performance indicator;explanation facility;Expert system;business rules;business intelligence | Issue Date: | 2012 | Publisher: | Univerzitet u Beogradu, Fakultet organizacionih nauka | Abstract: | Problem istraživanja doktorske disertacije se odnosi na veliku količinu podataka i nedostatak informacija u izveštajima sistema za poslovno izveštavanje. Iako uglavnom prepuni grafika i tabela, izveštaji sadrže samo podatke tj. njihove grafičke predstave koje bi trebalo da omoguće jednostavniji uvid u poslovanje. To dovodi do posledice da korisnici ovih izveštaja moraju ručno da tumače ove podatke (pregledanjem više izveštaja), pa mogu da propuste da uoče neke bitne informacije zbog: prevelike količine podataka, premorenosti, nedostatka koncentracije, nedostatka znanja ili iskustva, zbog subjektivnosti ili zlonamernosti. U radu se predlaže automatizacija procesa tumačenja podataka tj. njihovog pretvaranja u informacije. Ceo pristup se zasniva na relaciji koja definiše da podaci postaju informacije kada im se doda značenje (tumačenje). Da bi se dobilo tumačenje podataka i oni pretvorili u informacije, potrebno je odgovarajuće (domensko) znanje. Predložena realizacija ovog pristupa obuhvata primenu metoda, tehnika i tehnologija ES u cilju automatizacije procesa tumačenja vrednosti poslovnih pokazatelja. Konkretno, predlaže se da se deklarativni deo znanja (za tumačenje vrednosti pokazatelja) predstavi putem klasa, a proceduralni deo znanja putem pravila i fuzzy teorije. Zaključivanje se može izvršiti ulančavanjem unapred i fuzzy zaključivanjem, a informacije (zaključci) mogu da se pretvore u rečenice koje liče na govorni jezik korišćenjem tehnike učaurenog teksta. Posle napravljenog pregleda postojećeg stanja iz oblasti sistema za poslovno izveštavanje i ekspertnih sistema detaljno je definisan i obrazložen predloženi pristup sa teorijskog stanovišta. Diskusijom se ukazalo na (potencijalne) pozitivne i negativne aspekte rešenja, a zatim je dat i plan realizacije rešenja. Demonstracija pristupa putem izrade prototipa - PT aplikacije (PT - enterprise Profit interpretation Tutor) je omogućila proveru i dokazivanje postavljenih hipoteza. Izveštaji koje ova aplikacija pravi sadrže i podatke (u vidu grafika i tabela) ali i informacije u vidu rečenica kontrolisanog jezika... Dissertation research problem relates to the large amount of data and lack of information in the reports for business reporting systems. Although rich with charts and tables, reports actually contain only data (their graphical representations) and little or no information. Report users need to manually interpret this data (reviewing various reports) and, as consequence, may fail to detect some important information because of: too much data, fatigue, lack of concentration, lack of knowledge or experience, because of subjectivity, or malice. This paper proposes automating the data-to-information transformation process. The proposed approach is based on the relation that states that data becomes information when meaning (interpretation) is added to it. Also, appropriate (domain) knowledge is required in order to obtain meaning from data. The proposed implementation of this approach involves applying expert system methods, techniques and technologies in order to automate key performance indicator interpretation. Specifically, declarative knowledge is to be presented by classes, and procedural knowledge with rules and fuzzy theory. The information can be derived by forward chaining and fuzzy inference, and it can be then converted into natural-language-like sentences by using canned text. After reviewing the current situation in the field of business intelligence systems and expert systems, the proposed approach is explained in detail from the theoretical point of view. It is then followed by a discussion focusing on the (potential) positive and negative aspects of the solution, and then by the implementation plan. By demonstrating the approach with a prototype - PT application (PT - Enterprise Profit Interpretation Tutor), testing and proving of the hypotheses was enabled. Reports that the application provides contain data (in the form of graphics and tables) but also information (in the form of natural-language-like sentences). The evaluation consisted of an evaluation study in which subjects were students who used PT as a teaching aid. The results show that students believe that PT is useful and relatively easy to use. Explanations in the reports are considered useful and easy to understand, but would need to be more comprehensive. The largest number of students surveyed believe that working with PT application to some extent contributes to increasing their knowledge and that it provides a faster and somewhat more interesting way to mastering profit interpretation... |
URI: | https://nardus.mpn.gov.rs/handle/123456789/3056 http://eteze.bg.ac.rs/application/showtheses?thesesId=767 https://fedorabg.bg.ac.rs/fedora/get/o:7076/bdef:Content/download http://vbs.rs/scripts/cobiss?command=DISPLAY&base=70036&RID=515130778 https://rfos.fon.bg.ac.rs/handle/123456789/63 |
| Appears in Collections: | Doktorati |
Show full item record
Google ScholarTM
Check
This item is licensed under a Creative Commons License
