Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


27991.pdf (22.51 MB)

ETD Abstract Container

Abstract Header

MANILA: A Multi-Agent Framework for Emergent Associative Learning and Creativity in Social Networks

Shekfeh, Marwa
http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511861383686974

Abstract Details

2017, PhD, University of Cincinnati, Engineering and Applied Science: Computer Science and Engineering.
Since the rise of social media and the emergence of group discussions and blogs, the study of human communities has become both more important and more feasible, with data available through the Internet. Many aspects of human societies, such as the emergence of communities, city formation, and the evolution of languages, have become important topics of research. The research presented in this dissertation focuses on the study of one such phenomenon: Implicit collective learning and innovation in social systems. While explicit instruction is an important mode of learning, most learning in human systems occurs implicitly as an emergent result of the exchange of ideas among individuals. It is also mainly through such learning that innovative ideas can arise in individual minds by the recombination of received ideas. Thus, implicit collective learning is a very important facet of human life, the growth of human knowledge, and human creativity. The quality of such learning depends on many factors, including the quality of knowledge being exchanged, the preferences of individuals in terms of how they value information from various peers, the willingness of individuals to express their own thoughts, etc. Because this type of learning -- unlike explicit, goal-directed learning -- is very difficult to study experimentally in systems of sufficient size, it is important to understand it through computational modeling. The system proposed in this research is motivated by this need. The central contribution of the dissertation is the development and implementation of a comprehensive and powerful multi-agent framework for studying how the exchange of ideas among agents in a social network shapes the epistemes (knowledge bases) of individual agents under various scenarios. The framework -- the (Multi-Agent Network for the Implicit Learning of Associations) (MANILA) -- allows users to specify several aspects of the system, including: 1) The structure of the social network; 2) The propensity of agents to generate, express, and accept ideas; 3) The peer learning preferences of agents, i.e., which class of peers they are more willing to be influenced by; and 4) The quality and distribution of initial knowledge in the system. To make the simulations meaningful, the system provides a simple grounding mechanism in the form of an Oracle that can evaluate the objective correctness of expressed ideas and reward agents accordingly, thus creating a meritocratic labeling system that agents can potentially (but not necessarily) use to guide their learning. An important aspect of MANILA is the focus on associations, i.e., the connection between semantic elements (concepts), since associations are known to form the principal substrate of declarative knowledge. Ideas in the system are defined as combinations of associations between concepts, and represented as small semantic graphs.
Ali Minai, Ph.D. (Committee Chair)
Raj Bhatnagar, Ph.D. (Committee Member)
Karen Davis, Ph.D. (Committee Member)
Simona Doboli, Ph.D. (Committee Member)
Carla Purdy, Ph.D. (Committee Member)
254 p.
Computer Science
Multi-Agent Systems; Implicit Learning; Learning; Creativity; Social Networks

Recommended Citations

Citations

  • Shekfeh, M. (2017). MANILA: A Multi-Agent Framework for Emergent Associative Learning and Creativity in Social Networks [Doctoral dissertation, University of Cincinnati]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511861383686974

    APA Style (7th edition)

  • Shekfeh, Marwa. MANILA: A Multi-Agent Framework for Emergent Associative Learning and Creativity in Social Networks. 2017. University of Cincinnati, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511861383686974.

    MLA Style (8th edition)

  • Shekfeh, Marwa. "MANILA: A Multi-Agent Framework for Emergent Associative Learning and Creativity in Social Networks." Doctoral dissertation, University of Cincinnati, 2017. http://rave.ohiolink.edu/etdc/view?acc_num=ucin1511861383686974

    Chicago Manual of Style (17th edition)

ucin1511861383686974
135
© 2017, all rights reserved.
This open access ETD is published by University of Cincinnati and OhioLINK.