Steels, L. (1996b). Self-organising vocabularies. In Langton, C. G., & Shimohara, K. (Eds.), Artificial Life V. MIT Press, Cambridge, MA.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....links to nTLinks other peers. The peers share nConcepts concepts C 1 . C nConcepts , but use distinct names to refer to them. Thus we study the problem of peers sharing the same concepts but lacking knowledge of how to refer to them by names. This is somewhat similar to the approach taken in [25], without aiming at universally agreed upon names. Each peer p i uses its own set of names n pi to identify the concepts. We write (n pi ## C l ) when peer p i uses name n pi to refer to concept C l . These names may be seen in our data model, for example, as attribute names indicating the ....

....Thus their approach deals with the problem of having possibly contradictory knowledge on schema relationships. 18] propose an architecture for the use of XML based annotations in P2P systems to establish semantic interoperability. An approach to self organizing vocabularies is described in [25]. A set of agents communicate by randomly associating a fixed set of words to a fixed set of meanings (which is called a vocabulary but in fact is an ontology) and repeatedly evaluate how successful their communicative acts have been. Depending on the success the binding between a word and a ....

[Article contains additional citation context not shown here]

L. Steels. Self-organising vocabularies. In Proceedings of Artificial Life V, 1996.

....as a part of their experiment[12, 13] While this has provided a useful starting point, others argue that the use of a fixed lexicon is not representative of real world language development. Much research has been done focusing on a dynamic lexicon in a population of communicating organisms[8, 14, 15, 16, 17, 18]. The population in an experiment conducted by Hutchins and Hazlehurst[8] consists of neural networks with an additional structural feature. In addition to a hidden layer, the networks contain a hidden verbal input output layer. This extra layer performs the task of feature extraction used to ....

L. Steels. Self-organising vocabularies. In Proceedings of Artificial Life V, 1996.

....is shown how a prototypical form for members of a syntactic category may gradually emerge, thus making it easier to guess to which syntactic category an unknown form belongs. 1 Introduction There is a growing body of work exploring the idea that language can be viewed as a complex adaptive system [8], 4] The research explores formal models of language use in evolving inhomogeneous populations through computational simulations and experiments with physical robotic agents. Shared linguistic conventions have been shown to emerge as attractors of a language dynamics operating over autonomous ....

....parsing and generating. The resulting grammars generate purely formal structures but do not relate them with meaning. Subsequent work has focused on the problem how syntactic structure may be used to express meaning. Computational experiments in this line, particularly as reported by [5] 1] and [8], have shown that hierarchical structure starts to develop as soon as agents try to reuse in some way earlier created form meaning associations. This can either be because existing form meaning chunks are assembled in larger wholes, as emphasised in [8] or because an existing chunk is decomposed ....

[Article contains additional citation context not shown here]

Steels. L. (1996) Self-organising vocabularies. In Langton, C. (ed.) Proceedings of Artificial Life V. Nara, 1996.

....[13] Some models [2,10] have used neural networks and genetic algorithms for simulating the emergence of single word languages. For example, organisms controlled by neural networks evolve a shared lexicon of two signals for naming two different types of food sources [2] In other studies [4,12], agents use signal meaning matrices for communication games. At the beginning of evolution all signals are randomly associated to all possible meanings. During evolution adaptive pressure only strengthens associations between one signal and one meaning. Communication games have also been used for ....

Steels L. (1997). Self-organising vocabularies. In C.G. Langton & K. Shimohara (eds), Proceedings of the ALIFE V conference, Cambridge MA: MIT Press, 179-184.

....in the most adaptive pairs, the utterance, produced by a sighted but stationary female, causes the blind male to move in one of four directions such that it draws nearer to the female. Other work builds a self organizing system that converges onto a convention of word meaning associations (Steels, 1997). The universe of discourse is the set of agents themselves, each of which is uniquely defined by a finite set of meanings, or feature value pairs. Experiments described have on the order of twenty agents, five meanings, and five words. The agents build lexicons that associate the finite set of ....

Steels, L. (1997). Self-organising vocabularies. In (Langton and Shimohara, 1997), pages 136--141.

....Mechanisms In this paper, we regard language simply as combinations of words and meanings , namely vocabularies. Meanings take the form of objects. We adopt a language game model to form a language among agents. The model used here is a simplified one, which was firstly proposed by Steels [7]. The language formed on this model is open and adaptive in a sense that it may continue to function well even if new agents and new unknown objects are added [7] Furthermore, we investigate the consequences of contact between the different languages of distinct groups of agents. 2.1 Modelling ....

....a language game model to form a language among agents. The model used here is a simplified one, which was firstly proposed by Steels [7] The language formed on this model is open and adaptive in a sense that it may continue to function well even if new agents and new unknown objects are added [7]. Furthermore, we investigate the consequences of contact between the different languages of distinct groups of agents. 2.1 Modelling We define a set of communicative agents A = fa 1 ; am g and a set of objects O = fo 1 ; on g. A word w is expressed as a consonant vowel sequence, such ....

[Article contains additional citation context not shown here]

Luc Steels. Self-organising vocabularies. In Proceedings of Artificial Life V, 1996.

....be similar to the extent that the agents are in the same environment and use the same sensori motor apparatus. 4 Naming games for form meaning repertoires Other work has focused on how a shared set of form meaning relations could collectively be built up by a population of agents (see e.g. 5] [12]) Once again there is a population of agents. They play naming games. The agents have two components in their cognitive architecture: A mechanism for categorisation, such as the discrimination trees discussed earlier, and a lexicon which consists of a two way associative memory storing ....

Steels, L. Self-Organising Vocabularies. In: Langton, C. Proceedings of Alife V. The MIT Press, Cambridge. (1996)

....Sufficiently complex meaning then becomes the basis for other cognitive activities like planning, cooperation, or problem solving. We have already reported how agents may autonomously develop distinctions [19] and how they may develop autonomously a lexicon for expressing these distinctions [20]. A first experiment in physical grounding, in which these capabilities were instantiated on robotic agents playing adaptive language games, has been presented in [23] The present paper goes beyond this earlier work by using vision as source of sensory data and by showing the very beginnings of ....

....languages. Our goal is to understand the principles, not mimick human language genesis in toto. 4 Inventing and lexicalising distinctions This section focuses on mechanisms for sensory processing, meaning creation and lexicon formation. They have already been described in other papers [19] [20], 21] which should be consulted for a more extensive and formal discussion. The grammatical component is discussed in the next section. Ignoring for the time being grammar, the general architecture of the system built so far is as in figure 3. Each of these components is now discussed in some ....

Steels, L. (1996b) Self-organising vocabularies. In Langton, C. (ed.) Proceedings of Artificial Life V. Nara, 1996.

No context found.

Steels, L. (1996a) Self-organising vocabularies. In: Langton, C. (ed.) Proceedings of Artificial Life V. Nara.

.... our earlier work showing how a shared language medium in the form of a shared phonology may arise in a group of distributed agents [De Boer, 1997] how agents may autonomously develop distinctions [Steels, 1996a] and how they may develop autonomously a lexicon for expressing these distinctions [Steels, 1996b] A first experiment in physical grounding, in which these components were instantiated on robotic agents playing adaptive language games, has been reported in [Steels and Vogt, 1997] The present paper goes beyond this earlier work by showing the very beginnings of syntax. The rest of the paper ....

....4 Inventing and lexicalising distinctions The general architecture of the system built so far is as in figure 3. This section examines briefly the components for sensory processing, meaning creation and lexicon formation. They have been described already in other papers [Steels, 1996a] Steels, 1996b] which should be consulted for a more extensive and formal discussion of these mechanisms. The syntactic component is discussed in the next section. 4.1 Sensory Processing The tracking and image processing algorithms identify coherent image elements. For example, the robot moving around yields ....

[Article contains additional citation context not shown here]

Luc Steels. Self-organising vocabularies. In Langton, C. (ed.) Proceedings of Artificial Life V. Nara, 1996.

....autonomous. They acquire their own knowledge and decide for themselves how to communicate or divide up their world. 4. There is no global synchronisation. The system can operate in a fully distributed parallel fashion. The proposed principles have been implemented in software simulations [6] 7] [8] and have been integrated in robotic agents, in which case the ontology is based on an embodied physical interaction with the environment [12] 9] This research is strongly related to a growing body of work on the origins of (natural) languages, extensively reviewed in [10] The rest of the ....

Steels, L. (1996c) Self-organising vocabularies. In Langton, C. (ed.) Proceedings of Artificial Life V. Nara,1996. The MIT Press, Cambridge Ma.

No context found.

Steels, L. (1996b). Self-organising vocabularies. In Langton, C. G., & Shimohara, K. (Eds.), Artificial Life V. MIT Press, Cambridge, MA.

No context found.

Luc Steels. Self-organising vocabularies. In Christopher G. Langton and Taksunori Shimohara, editors, Proceedings of Artificial Life V. MIT Press, 1996b.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC