Razporejanje predavanj na konferenci

AVDIČ, EMIL (author), Brodnik, Andrej (mentor), Papa, Gregor (co-mentor)

V nalogi bomo predstavili nekaj algoritmov, ki jih je mogoče uporabiti za izdelavo urnika konference. Najprej predstavimo t.i. požrešni algoritem. Temu sledijo še zahtevnejši algoritmi, ki izhajajo iz problema pokritja in problema nahrbtnika. Vse tri algoritme testiramo na treh, različno zgrajenih konferencah, s pripadajočimi članki. Prvo testno konferenco smo kreirali povsem poljubno z izmišljenimi podatki. Druga testna konferenca je bila psevdo naključna, ki jo program zgenerira sam glede na parametre, ki jih podamo. Tretja je konferenca, kjer podatke dobimo iz dejanskih preteklih konferenc ali iz prihajajoče konference. Izbrane nabore člankov znotraj testnih konferenc smo uredili v seje s pomočjo treh omenjenih algoritmov. Oceno urejanja člankov konference smo ocenili preko izračuna kromatičnosti in praznih prostorov v seji. Glede na naše kriterije ocenjevanja se za najboljšega izkaže algoritem s hevristiko nahrbtnika, ki izhaja iz algoritma prvega prileganja. Na koncu predstavimo rezultate primerjave.

• Language: Slovenian
• Keywords: generiranje urnika, NP-problem, problem nahrbtnika, prolem pokritja, požrešni algoritem
• Work type: Bachelor thesis/paper
• Organization: FRI - Faculty of Computer and Information Science
• Year: 2016
• PID: 20.500.12556/RUL-81217
• Publication date in RUL: 07.04.2016

Secondary language

• Language: English
• Title: Talk scheduling at the conference

Abstract

We will be introducing a few algorithms, that can be used for creating a schedule of a conference. Firstly we introduce greedy algorithm. We follow up with a few more complicated algorithms, that originate from the bin packing problem and the set cover problem. All three algorithms are then tested on three differently build conferences with their corresponding articles. First conference is created randomly with fictional data. Second conference is created with a pseudo random function. Third conference has real life data from a conference that is yet to come. Selected sets articles are sorted with the help of previously mentioned algorithms. Quality of algorithms is measured through a function of chromaticity and empty spaces in sessions. According to our criteria, the best algorithm is the one with the heuristics of the bin packing problem, coming from the first fit algorithm. At the end we present the results of testing.

• Keywords: generating schedule, NP-problem, backsack problem, set cover problem, greedy algorithm