A Dynamic-Oriented Decision Support System for Group Interview Knapsack Problem

• Published in: Augmented and Virtual Reality pp. 498 - 506
• Main Authors: Jouida, Sihem Ben, Krichen, Saoussen
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing 2015
• Series: Lecture Notes in Computer Science
• Subjects: Decision support system; Dynamic programming; Online knapsack problem; Optimal stopping
• ISBN: 3319228870; 9783319228877
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-22888-4_39

We address in this paper a dynamic platform for knapsack selection that handles sequential arriving items looking for an optimized investment in terms of values and risk aversion. Investment items, specified by their costs and their values (e.g. stochastic modelling of the possible rewards that can be generated), arrive sequentially, by groups, over time to be later scheduled for an eventual investment. It is worth mentioning that items arrive over time to be firstly evaluated, then accepted or discarded, based on partial previous information about already observed items and no information about forthcoming ones. Such decision depends solely on the decision makers ranking of the arriving items. The decision support system (DSS) inputs the calibration of the decision makers preference levels and the whole set of problem parameters followed by their probability distributions. From a theoretical point of view, this problem can be viewed as an online knapsack problem, an NP-hard optimization problem solved using a dynamic programming algorithm. The proposed platform is experienced on numerous problem instances to show its effectiveness in generating profitable decisions.