Full-text

URL: http://arti.vub.ac.be/~joris/publication...xicon-experiments.ps

Related links

Web search: Google Web Search   ::   Google Scholar

Within this site: References (13)

Paper at a Glance

Robotic Experiments on
the Emergence of a Lexicon
Joris Van Looveren
Artificial Intelligence Laboratory
Vrije Universiteit Brussel
Pleinlaan 2, 1050 Brussels, Belgium
e­mail: joris@arti.vub.ac.be

Abstract The Talking Heads experiment is a robotic version of earlier experiments on the emergence of a lexicon. A speaker and a hearer agent chosen ran­ domly from a larger population try to communicate with each other about objects they see in their environment; the hearer's task is to guess the topic that the speaker talks about. The agents use bodies equipped with pan­tilt cameras to perceive their environment, which consists of a whiteboard con­ taining geometrical shapes that form the topics of the interactions between the agents. The paper examines data gathered during a four­month run involving several interconnected Talking Heads sites.
1 Introduction The lexicon is a very important part of language; without it, complex communi­ cation would not be possible. In its simplest form, it can be considered to be a list of associations between symbols and representations of concepts. The symbols are essentially arbitrary; the majority of words we know have no direct relationship with the concepts they represent. A consequence of the arbitrariness is that, when a child grows up and learns its native language, it has to learn all associations from scratch, without any prior knowledge. An important question is then how these associations can be learned. Many experiments have already been done in the area of lexicon formation, but most of these were done in simulation, using the game paradigm from economics and game theory. This paper will describe an experiment in which a mechanism for lexicon cre­ ation and learning (which has been tested in simulations before) has been imple­ mented on robotic bodies, that live in a real­world environment, using the sensors present in the body. 2 Language Models Many linguistic theories implicitly assume a single, ideal speaker; they do ...

@inproceedings{vanlooveren01roboticExperiments,
  author={J. Van Looveren},
  title={Robotic Experiments on the Emergence of a Lexicon},
  year={2001},
  address={Amsterdam, The Netherlands},
  editor={Krose, B. and et al.},
  booktitle={Proceedings of the 13th Belgium-Netherlands Conference on Artificial Intelligence (BNAIC'01)},
  note={Best original paper award},
  url={http://groups.lis.illinois.edu/amag/langev/paper/vanlooveren01roboticExperiments.html}
}

HOME   ::  Conference List   ::   Conference Paper	Comments to: junwang4 you-know-at gmail.com	Last update: 6/16/13