Abstract

Everyone belongs to social categories based on factors such as musical preference, fashion, gender, or race. Yet, some category members are more central, privileged, or marginalized than others. Membership in such social categories is also dynamic: whether someone is a member or not may change over time, both within and between groups. Chimeria is a system to help people better understand social categorization phenomena such as marginalization and the dynamics of group membership. Chimeria does this through an interactive narrative. Consider the following story on a music-oriented social network:

A punk rock music fan decides to listen to a little jazz. She listens to a couple of albums by the jazz musician Thelonious Monk, but however, still continues to post messages only about punk rock. She grows tired of being a punk rocker (who dabbles in jazz on the side), so listens again only to hardcore punk rock music. But now, upon returning back to punk rock, punk rock seems to have lost its luster. She finally decides to forsake punk rock and become a jazz fan…for good.

In this story, a central member of a category moves toward the margins of the category, back toward the center, and finally changes categories altogether at the end. This pattern of movement within and between categories is the sort of phenomenon that Chimeria models mathematically. The story could have been one of racial or gender group membership; but for the initial version of Chimeria we have chosen music-related identity as a focus domain.

These dynamics of group membership are simulated in Chimeria using an artificial intelligence (AI)-driven interactive story. Users interact with various characters within a novel social networking interface. Based on a player’s musical preferences and decisions in an online conversation, the system dynamically generates an interactive conversation centered upon snippets of music that takes place between the user and the other characters. This interactive conversation, grounded in a sociolinguistics model of conversational narrative, allows an engaging experience in which players may encounter various social category membership phenomena which occur in the real-world.

Most computational and interactive systems, which use traditional AI categorization approaches, often focus only on developing algorithms for optimal classification into discrete, static boxes. Yet, membership in social categories is gradient: some members are more central or marginalized than others. Membership in such social categories is also dynamic: the statuses of members may change over time, both within and between groups. Systems such as computer games and social networks represent social categories in forms such as character classes, character races, groups based on privacy settings, etc. that currently are seldom gradient or dynamic. Chimeria computationally models gradient and dynamic social category membership phenomena typically addressed in social and cognitive sciences, such as partial, multiple, “authentic,” or feigned group memberships.

Narratives are often used to form, convey, and reinforce memberships in social groups, and thus, a robust model of group membership can be an important aspect for modeling many everyday forms of narrative. The core of Chimeria, the Chimeria Engine, implements a model of narratives of group membership in social media through the Chimeria Social Narrative Interface (Chimeria-SN). This is useful both for generating expressive narrative content, and for the analysis of social categorization phenomena. This component is grounded in a sociolinguistic model of conversational narrative, with some adaptations and extensions in order for it to be applied to an interactive social networking domain. One eventual goal of this work is to be able to extrapolate social group membership by analyzing narratives in social networks; this paper deals with the inverse of that problem, namely, synthesizing narratives from a model of social group membership dynamics.

PUBLICATIONS

D. Fox Harrell, Dominic Kao, Chong-U Lim. “Computationally Modeling Narratives of Social Group Membership with the Chimeria System,” Proceedings of the 2013 Workshop on Computational Models of Narrative – a satellite workshop of CogSci 2013: The 35th meeting of the Cognitive Science Society, Berlin, Germany, July 31 – August 3. [PDF]

Associated Project: 
Launch application