Use this URL to cite or link to this record in EThOS:
Title: Acquiring and grounding a lexicon with art : towards robotic systems that understand language
Author: Chandler, Nathan James
ISNI:       0000 0001 3527 3234
Awarding Body: Nottingham Trent University
Current Institution: Nottingham Trent University
Date of Award: 2000
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
This thesis reports on the development of a method to enable artificial systems, specifically robots, to autonomously acquire a diverse lexicon of natural language terms. In this case English words which can be used to facilitate communication between themselves and their human operators. For reasons that will become clear, this task is referred to as 'lexical acquisition and grounding.' It is argued herein that in order for this communication to succeed, the artificial systems must gain an intrinsic understanding of the meaning of these words. Moreover, it is argued that such an intrinsic understanding is ultimately founded upon the systems ability to experience, via sensory and motor systems, its surrounding environment. Put in simple terms, this is similar to arguing that a blind man could never truly understand an English word such as 'green.' Support for these arguments is provided by way of a comprehensive analysis and interpretation of a variety of recent theories (e.g. Hamad's Symbol Grounding Theory) from Cognitive Science, Psychology, and Philosophy which reject a number of 'traditional' approaches to defining the meanings of words on various grounds. A review of related work deals with a number of practical models that have been developed from similar theoretical foundations. Based upon an assessment of the strengths and weaknesses of each approach an alternative approach to the lexical acquisition and grounding task is suggested. This approach applies an existing theory of computation - Adaptive Resonance Theory (ART) - to this task. This new application of ART is examined empirically and then ART models are adapted and extended in relation to various task specific requirements. The aim and objectives of the work described in this thesis have been achieved, in that a model has been developed that is able to autonomously acquire English words in an incremental and continual manner and demonstrates that it has an understanding of these words. Moreover, it has been shown that this model can concurrently learn meanings with respect to more than one sensory domain and that the model can operate in a manner that is fast enough to warrant its practical implementation. With support from empirical investigation and theoretical analysis it has been shown that, on several counts, the application of an ART-based framework to the lexical acquisition and grounding task provides significant advance on systems that have been developed in the past to address this problem.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: Symbol; Semantics