Composing agents from role prototypes of social groups

From Master Projects
Jump to: navigation, search


has title::Composing agents from role prototypes of social groups
status: finished
Master: project within::Computational Intelligence and Selforganisation
Student name: student name::Jeroen de Man
Dates
Start start date:=2011/02/01
End end date:=2011/08/15
Supervision
Supervisor: Tibor Bosse
Second reader: has second reader::Jurriaan van Diggelen
Company: has company::TNO
Poster: has poster::Media:jmn300_poster.pdf

Signature supervisor



..................................

Abstract

Project Thesis

"The individual has as many social egos as there are different social groups and strata with which he is connected." (Sorokin, 1962)

Culture, religion, work, family or a sports club are only a few examples of factors defining social groups. Each of such groups has its own norms and goals which members are expected to adhere to. Therefore, it is not surprising that an individual is shaped by the social groups (s)he is connected with, required to balance the expectations of one group with those of another. Go to a football game or the church? Work late or go home? Although cognitive agents are created for a large variety of applications, they are not often required to cope with such dilemmas. Or, as Pokahr et al. (2005) put it, `the cumbersome task of ensuring that the agent will never process any conflicting goals at the same time is left to the agent developer'.

We are interested in an approach where the agent itself can decide between goals. We even take it one step further by aiming at a situation in which social groups can be modeled independent from each other and an agent connected to multiple groups is still able to decide which goal to pursue. We will use the BDI structure (Rao and Georgeff, 1995) as a base and investigate how a social group can be modeled as a separate set of beliefs, goals and plans. Unfortunately, just merging the BDI modules is not possible as numerous conflicts can occur, thus we need to find a way to decide which actions to take regardless of any conflicts. We will investigate a way to express a quantitative rating of plans, inspired by the Culturally-Affected Behavior (CAB) prototype (Solomon et al., 2008) and how it is used in the culturally and emotionally affected behavior model of Bulitko et al. (2008), such that we simply can select the plan with the highest rating gets selected.


KIM 1 Slides

"The individual has as many social egos as there are different social groups and strata with which he is connected." (Sorokin, 1962)

Culture, religion, work, family or a sports club are only a few examples of factors defining social groups. Each of such groups has its own norms and goals which members are expected to adhere to. Therefore, it is not surprising that an individual is shaped by the social groups (s)he is connected with, required to balance the expectations of one group with those of another. In this presentation, I will show why modeling social groups is useful in the context of simulation-based military training. After that, a short overview is given of the Culturally-Affected Behavior (CAB) prototype (Solomon et al., 2008) and how it is used in the culturally and emotionally affected behavior model of Bulitko et al. (2008). Inspired on these models, an idea to model social groups is given as well as a number of research questions, expected problems and desired goals of the project.

KIM 2 Slides

"The individual has as many social egos as there are different social groups and strata with which he is connected." (Sorokin, 1962)

Culture, religion, work, family or a sports club are only a few examples of factors defining social groups. Each of such groups has its own norms and goals which members are expected to adhere to. Therefore, it is not surprising that an individual is shaped by the social groups (s)he is connected with, required to balance the expectations of one group with those of another. In the past months, we have developed a model which is able to combine several social groups implemented as BDI modules into a single agent and cope with conflicts arising from these different groups. During this presentation, I will explain this model in detail, cover the methods used to guarantee consistent behavior and show a small example relating to a military training scenario. This example will show different behavior is generated when an agent is connected with different social groups and that a single agent can act differently when placed in another context.

References

Bulitko, V., Solomon, S., Gratch, J., and Lent, M. v. (2008). Modeling culturally and emotianally affected behavior. In Proceedings of the Fourth Artificial Intelligence and Interactive Digital Entertainment Conference, pages 10-15, Stanford, California, US.

Pokahr, A., Braubach, L., and Lamersdorf, W. (2005). A goal deliberation strategy for BDI agent systems. In Third German conference on Multi-Agent System TEchnologieS (MATES-2005), pages 82-94. Springer-Verlag.

Rao, A. and Georgeff, P. (1995). BDI agents: From theory to practice. In Proceedings of the first international conference on multi-agent systems (ICMAS-95), pages 312-319.

Solomon, S., Lent, M. v., Core, M., Carpenter, M., and Rosenberg, M. (2008). A language for modeling cultural norms, biases and stereotypes for human behavior models. In Proceedings of the 17th Conference on Behavior Representation in Modeling and Simulation (BRIMS 2008).

Sorokin, P. (1962). Society, Culture and Personality: Their Structure and Dynamics. Cooper Square Publishers, New York, US.