Μελέτη συστήματος συστάσεων από ομάδες χρηστών με βάση την δικαιοσύνη
View/
Keywords
Recommendations ; Συστημάτων συστάσεωνAbstract
The present master thesis focus on the computational optimization of a special category of recommendations systems. In particular, the objective is to create a system, based on a greedy algorithm, that addresses efficiently the complex problem of recommending packages of objects to user groups which are constrained by the fairness between all the members of the group. The “fairness” term derives from the fact that the proposed objects must meet a specific criterion among the members of the group and the algorithm aims on the “smart” selection of those. Moreover, an introduction is made to the theory of recommendations systems, the different implementation approaches that exist, what problems they are trying to solve, the use of constrains on such systems and the effect these limitations have on their performance. Furthermore, a detailed description is presented of the approximation algorithm that was developed for generating fairness constrained based recommendations where the generated results are evaluated with the equivalent complete brute force solutions. Finally, all the parameters that affect the performance of the algorithm are analyzed and conclusions can be drawn that can be used for future research.