Thesis subject

Creating socially intelligent virtual agents

Level: MSc

Research area/discipline: Computational Social Simulation, Agent-based modelling, Artificial sociality

Prerequisites: Programming in Python (INF-22306) and/or Software Engineering (INF-32306) and/or Agent Based Modelling (INF-50806)

Short description:

Virtual agents are starting to be used in all kinds of applications, e.g. in industry and in training. Such agents need to understand the social world. The INF group developed a framework (GRASP: Groups, Rituals, Affiliation, Status and Power) for the social intelligence of agents. What is needed is a convincing application. 
The first application will be a GRASP version of the existing Playground model. In this model, school children form groups and friendships, or they quarrel, and in so doing they build a network of status and popularity. This version will be built in Netlogo. 
The existing playground model has affiliation, status and power but very limited group and ritual features. These will need to be improved.

Reference: Hofstede, G. J., et al. (2015). Gender differences: the role of nature, nurture, social identity and self-organization. Multi-Agent-Based Systems Workshop (MABS) 2014, LNAI 9002. F. Grimaldo and E. Norling, Springer: 72-87.

For more information: