The socio-technical teams formation problem: Complexity, Mathematical Formulations and Computational Results

• Autor(a) principal: Tatiane Fernandes Figueiredo
• Data de Publicação: 2014
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UFC
• Texto Completo: http://www.teses.ufc.br/tde_busca/arquivo.php?codArquivo=15392
• topic: Trabalho cooperativo; GerÃncia de pessoas; Complexidade computacional; ProgramaÃÃo inteira; CombinatÃria poliÃdrica; HeurÃstica; ProgramaÃÃo heuristica; AdministraÃÃo de pessoal; Cooperative work; People manegement; Computational complexity; Integer programming; Polyhedral combinatorics; Heuristic; CIENCIA DA COMPUTACAO

Using concepts of the socio-technical systems theory, this dissertation defines mathematically the problems of cooperative teams formation considering social and technical constraints separately, and then presents their computational complexity. Mainly, it is defined and studied the central problem in this work, which jointly considers social and technical requirements for creating teams of cooperative work, to be called FEST (Socio-Technical Teams Formation Problem). Two mathematical formulations and a meta-heuristic are proposed for FEST. One formulation uses a cubic number of variables and constraints, whereas the second one has a quadratic number of variables but an exponential number of constraints. The proposed heuristic is based on the Non-monotonic Simulated Annealing meta-heuristic with local search using swap-like operators. The correctness of both formulations is proved. A polynomial algorithm to separate the constraints of the second formulation is presented. It is proved that the two formulations provide the same linear programming bound, and valid inequalities to strengthen it are proposed. For the compact formulation, some classes of valid inequalities are shown to be facet-inducing under suitable hypotheses. Finally, it is statistically analyzed the performance of the presented formulations and meta-heuristic. Real and random generated instances are used in the computational experiments.