Stand H3

About the Project

The main idea behind the project is to build a multiagent system that can select the teams of employees for different roles in a particular workplace. Each role would have a budget that should not be exceeded. This problem will be broken down into a series of knapsack problems and each knapsack problem will be assigned to an agent. The multiagent would need to ensure that none of the employees are assigned to multiple roles. The agents solved the knapsack problems in parallel and used backtracking to ensure the none of the employees are assigned to multiple roles. The multiagent system had agents that made use of simulated annealing, search algorithms, genetic algorithms, and dynamic programming. The Multiagent system consisted of three pipelines. The first pipeline was a homogenous multiagent system whose agents made use of simulated annealing. The second pipeline was a homogenous multiagent system that made use of the genetic algorithm. The third pipeline was a diverse multiagent system that consisted of various types of agents that made use of genetic algorithms, simulated annealing, search algorithms and dynamic programming. The multiagent system’s ability to select candidates for the Analyst, Manager, Director, Developer, Business Analyst and Data Analyst roles with budgets of R400 000, R300 000, R500 000, R200 000, R500 000 and R800 000, respectively, was evaluated. The second pipeline achieved the highest rewards of 245.5, the third achieved the second highest rewards of 225 and the first pipeline which is a homogenous multiagent system that made use of simulated annealing per-formed the worst with rewards of 122.