Filtern
Dokumenttyp
Sprache
- Englisch (11)
Volltext vorhanden
- nein (11)
Gehört zur Bibliographie
- ja (11)
Schlagworte
- Gruppendynamik (3)
- Kollaboration (3)
- Computational social science (2)
- GitHub (2)
- Künstliche Intelligenz (2)
- Maschinelles Lernen (2)
- Soziologie (2)
- Affect (1)
- Affekt (1)
- Analytische Soziologie (1)
Recent advances in artificial intelligence and computer science can be used by social scientists in their study of groups and teams. Here, we explain how developments in machine learning and simulations with artificially intelligent agents can help group and team scholars to overcome two major problems they face when studying group dynamics. First, because empirical research on groups relies on manual coding, it is hard to study groups in large numbers (the scaling problem). Second, conventional statistical methods in behavioral science often fail to capture the nonlinear interaction dynamics occurring in small groups (the dynamics problem). Machine learning helps to address the scaling problem, as massive computing power can be harnessed to multiply manual codings of group interactions. Computer simulations with artificially intelligent agents help to address the dynamics problem by implementing social psychological theory in data-generating algorithms that allow for sophisticated statements and tests of theory. We describe an ongoing research project aimed at computational analysis of virtual software development teams.
Friend or Foe: A Review and Synthesis of Computational Models of the Identity Labeling Problem
(2021)
We introduce the identity labeling problem – given an individual in a social situation, can we predict what identity(ies) they will be labeled with by someone else? This problem remains a theoretical gap and methodological challenge, evidenced by the fact that models of social-cognition often sidestep the issue by treating identities as already known. We build on insights from existing models to develop a new framework, entitled Latent Cognitive Social Spaces, that can incorporate multiple social cues including sentiment information, socio-demographic characteristics, and institutional associations to estimate the most culturally expected identity. We apply our model to data collected in two vignette experiments, finding that it predicts identity labeling choices of participants with a mean absolute error of 10.9%, a 100% improvement over previous models based on parallel constraint satisfaction and affect control theory.
We review Affect Control Theory (ACT) as a promising basis for equipping computational agents in social simulations with a sense of sociality. ACT is a computational theory that integrates sociological insights about the symbolic construction of the social order with psychological knowledge about cognitive-affective mechanisms. After explaining the theoretical foundations of ACT and applications of the theory at the dyadic and group level, we describe a case study applying the theory from an ongoing research project examining self-organized online collaboration in software development.
A topology of groups
(2020)
In this work, we study the collaboration patterns of open source software projects on GitHub by analyzing the pull request submissions and acceptances of repositories. We develop a group typology based on the structural properties of the corresponding directed graphs, and analyze how the topology is connected to the repositorys collective identity, hierarchy, productivity, popularity, resilience and stability. These analyses indicate significant differences between group types and thereby provide valuable insights on how to effectively organize collaborative software development. Identifying the mechanisms that underlie self-organized collaboration on digital platforms is important not just to better understand open source software development but also all other decentralized and digital work environments, a setting widely regarded as a key feature of the future work place.
This paper studies the dynamics of identity and status management within groups in collaborative settings. We present an agent-based simulation model for group interaction rooted in social psychological theory. The model integrates affect control theory with networked interaction structures and sequential behavior protocols as they are often encountered in task groups. By expressing status hierarchy through network structure, we build a bridge between expectation states theory and affect control theory, and are able to reproduce central results from the expectation states research program in sociological social psychology. Furthermore, we demonstrate how the model can be applied to analyze specialized task groups or sub-cultural domains by combining it with empirical data sources. As an example, we simulate groups of open-source software developers and analyze how cultural expectations influence the occupancy of high status positions in these groups.
Belief systems matter for all kinds of human social interaction. People have individual cognitions and feelings concerning processes in their environment, which is why they may evaluate them differently. Belief systems can be visualized with cognitive-affective maps (CAMs; as reported by Thagard (in: McGregor (ed) EMPATHICA: A computer support system with visual representations for cognitive-affective mapping, AAAI Press, CA, 2010)). However, it is unclear whether CAMs can be constructed in an intersubjective way by different researchers attempting to map the beliefs of a third party based on qualitative text data. To scrutinize this question, we combined qualitative strategies and quantitative methods of text and network analysis in a case study examining belief networks about participation. Our data set consists of 10 sets of two empirical CAMs: the first CAM was created based on participants’ freely associating concepts related to participation in education (N = 10), the second one was created based on given text data which the participants represented as a CAM following a standardized instruction manual (N = 10). Both CAM-types were compared along three dimensions of similarity (network similarity, concept association similarity, affective similarity). On all dimensions of similarity, there was substantially higher intersubjective agreement in the text-based CAMs than in the free CAMs, supporting the viability of cognitive affective mapping as an intersubjective research method for studying the emotional coherence of belief systems and discursive knowledge. In addition, this study highlights the potential for identifying group-level differences based on how participants associate concepts.
Theoretical and Empirical Modeling of Identity and Sentiments in Collaborative Groups (THEMIS.COG) was an interdisciplinary research collaboration of computer scientists and social scientists from the University of Waterloo (Canada), Potsdam University of Applied Sciences (Germany), and Dartmouth College (USA). This white paper summarizes the results of our research at the end of the grant term. Funded by the Trans-Atlantic Platform’s Digging Into Data initiative, the project aimed at theoretical and empirical modeling of identity and sentiments in collaborative groups. Understanding the social forces behind self-organized collaboration is important because technological and social innovations are increasingly generated through informal, distributed processes of collaboration, rather than in formal organizational hierarchies or through market forces. Our work used a data-driven approach to explore the social psychological mechanisms that motivate such collaborations and determine their success or failure. We focused on the example of GitHub, the world’s current largest digital platform for open, collaborative software development. In contrast to most, purely inductive contemporary approaches leveraging computational techniques for social science, THEMIS.COG followed a deductive, theory-driven approach. We capitalized on affect control theory, a mathematically formalized theory of symbolic interaction originated by sociologist David R. Heise and further advanced in previous work by some of the THEMIS.COG collaborators, among others. Affect control theory states that people control their social behaviours by intuitively attempting to verify culturally shared feelings about identities, social roles, and behaviour settings. From this principle, implemented in computational simulation models, precise predictions about group dynamics can be derived. It was the goal of THEMIS.COG to adapt and apply this approach to study the GitHub collaboration ecosystem through a symbolic interactionist lens. The project contributed substantially to the novel endeavor of theory development in social science based on large amounts of naturally occurring digital data.
Missing data is a common, difficult problem for network studies. Unfortunately, there are few clear guidelines about what a researcher should do when faced with incomplete information. We take up this problem in the third paper of a three-paper series on missing network data. Here, we compare the performance of different imputation methods across a wide range of circumstances characterized in terms of measures, networks and missing data types. We consider a number of imputation methods, going from simple imputation to more complex model-based approaches. Overall, we find that listwise deletion is almost always the worst option, while choosing the best strategy can be difficult, as it depends on the type of missing data, the type of network and the measure of interest. We end the paper by offering a set of practical outputs that researchers can use to identify the best imputation choice for their particular research setting.