Resumen de Artificial intelligence methods to support people management in organizations
Organisations have shifted from work arranged around individual jobs to team-based work structures. A new generation of solutions for organisations must give support to team management by encouraging team effectiveness and introducing automation. In this dissertation, we tackle several different problems that are connected to team management in organisations. In particular, we contribute by proposing a people management workflow that addresses the problems connected to team composition as well as problems of accurate employee evaluation and task performance evaluation.
First, we review the literature on team composition and formation from both the organisational psychology and computer science perspectives and we explore the connection between individuals’ attributes and team performance as well as the cross fertilization opportunities between those fields.
Second, we review the most prominent tools to measure individuals' attributes, as these measures are necessary inputs for team composition processes. In particular, we describe the dominant approaches in Organisational Psychology, Industrial Psychology and Human Resources and summarise they main findings to measure individual personality and competences.
Third, we use our findings to propose a model to predict team performance given a task and based on individuals' attributes (i.e. competences, personality and gender). We define the Synergistic Team Composition Problem (STCP) as the problem of finding a team partition constrained by size so that each team, and the whole partition of employees into teams, is balanced in terms of individuals' competences, personality and gender. We propose two different algorithms to solve this problem: an optimal algorithm called STCPSolver that is effective for small instances of the problem, and an approximate algorithm called SynTeam that provides high-quality, but not necessarily optimal solutions. We present empirical results that we obtained when analysing student performance. Our results show the benefits of a more informed team composition that exploits individuals' competences, personalities and gender.
Fourth, we devise an algorithm called Collaborative Judgment (CJ) to fairly evaluate individuals' and teams' outcomes once tasks are performed. In particular, we want to diminish the importance of biases in the evaluation process by allowing evaluators to assess their peers, namely other evalutors. Our empirical results show the benefits of more informed assessment aggregation method.
