20.500.12556/RUL-95567 E-učni sistemi in priporočanje učnih gradiv E-learning systems and recommending learning materials V okviru našega dela smo razvili preprost e-učni sistem, ki priporoča gradiva dijakom. Najprej smo naredili kratek pregled o tem, kaj naj bi vsebovali e-učni sistemi. V nadaljevanju smo se lotili implementacije lastnega sistema, v katerem je na voljo tudi sprotna učna analitika. Za potrebe testiranja sistema smo naredili tudi nekaj gradiv, ki pokrivajo predmet informatika v gimnazijah. Potem smo v sistemu testirali tri različne načine priporočanja, in sicer: čim bolj podobno gradivo, čim bolj drugačno gradivo in naključni vrstni red gradiv. Vsakega od načinov je testirala tretjina dijakov. Na osnovi primerjave preciznosti, priklica in mere F1 smo ugotovili, da sta najboljša načina naključni in, takoj za njim, matrična faktorizacija, ki vrača medseboj čim bolj drugačno gradivo. Način, ki dijaku ponudi čim bolj medseboj podobno gradivo, se je v poskusih izkazal kot slab. Rezultati kažejo, da moramo učencu priporočiti naključno gradivo. Glede na preciznost z ostalimi načini priporočanja dosežemo slabše rezultate. Within our thesis, we developed a simple e-learning system which recommends learning materials to high school students. We began by preparing a short overview of appropriate e-curriculums. We continued by implementing our own system that includes real-time learning analytics. To test the system, we also created some learning materials on high school information science. We tested three different recommendation techniques, i.e. learning materials that are the closest match to previous materials and the furthest match from previous materials and materials in random order. Each material assignment strategy was tested on one third of students. Based on observing precision, recall and F1 measure, we established that the best system turns out to be random order material assignment, followed by matrix factorisation that offers the least matching materials to previous materials. The technique offering the closest matching materials proved to be the least useful. The results therefore indicate that we should offer students random materials. Other strategies provided considerably worse results when we compared them by precision. e-učni sistemi priporočanje gradiv učna analitika e-learning system recommendation of learning material learning analytics true false false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2017-09-20 20:04:43 2017-09-20 20:04:47 2022-08-10 06:32:47 0000-00-00 00:00:00 2017 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 1970-01-01 20741 Bavcon_David_-_E-ucni_sistemi_in_priporocanje_ucnih_gradiv.pdf Bavcon_David_-_E-ucni_sistemi_in_priporocanje_ucnih_gradiv.pdf 1 F387A2EC076E38E7C1D4F5D8A35CBE41 abcdf5881d3df52867f31ebb6f2765fd65185abb9871129097722a1291c69dea c9ff82d6-a1b3-11eb-a523-00155dcfd717 20.500.12556/rul/45e47253-1725-41ba-aee9-f839f3654c77 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=104176 Fakulteta za računalništvo in informatiko 0 0 0