Projects
Computational Social Science
Today’s Internet users actively distribute and share information via online communication in ways distinct from those of the real world. Nevertheless, online social interaction has the potential to change the real world.
We use Twitter, a popular form of social media, to observe and quantify online social interaction. Twitter users post short text messages, or ‘tweets,’ each consisting of no more than 140 characters and addressing what is happening now. Some tweets may immediately go viral over user networks via chains of replies and retweets. We study the real-time and networked nature of such online communication dynamics, as well as the evolution of the resultant information ecosystem on Twitter. Moreover, we develop a new method to efficiently quantify collective behavior from a large amount of social data. From these findings we model and simulate online social dynamics.
ALIFE modeling for the Evolution of Communication
Since communication is a behavior that engages dynamic interaction via media, it evolves according to changes in media formats and socio-technological environments. To reveal the dynamic and open-ended nature of communication, the analysis of “communication as it is” does not suffice. Artificial life (ALIFE) research therefore takes a constructive approach in which “communication as it could be” is modeled and simulated using computers, thereby improving our understanding of communication’s essential characteristics.
From the perspective that communication as a complex system, we investigate the evolutionary dynamics of linguistic communication in humans and pre-linguistic communication in animals via computer simulations, as well as explore some design principles for possible forms of communication (e.g., human–robot communication).
Modeling Action Grammars in Animals
The hypothesis that human language and animal communication can be compared in terms of computational ability is another watershed in studies of language evolution. Here, the most important ideas are perhaps the “faculty of language in a broad sense” (FLB) and the “faculty of language in a narrow sense” (FLN). While FLB is a cognitive ability shared by both humans and animals, FLN is a computational ability that allows recursive hierarchical operations, or ‘recursion,’ which are thought to be unique to humans. Since this hypothesis emerged, researchers have sought the precursor of recursion in animal behavior.
We study the structure of action grammar by comparing the communication systems of different species to reveal computational abilities that enable complex animal behaviors, including recursion. In so doing we develop efficient methods for analyzing action grammar, as well as construct a framework for understanding animal behavior in a unified manner.