identifying archaeological knowledge using multi dimensional scaling and multiple constraint satisfaction

Abstract

In this thesis, I look at the current state of research in two fields: the cognitive psychology of learning and expertise & the development of Intelligent Tutoring Systems, especially their methods of modelling the users knowledge state. Within these areas I proceed to examine the way that these theories have overlapped in the past and consider their recent divergence, suggesting that this parting of the ways is premature. I then consider other relevent research so as to suggest a hypothesis where a symbolic connectionist approach to the modelling of knowledge states could be a solution to previous difficulties in the field of Intelligent Tutoring. This hypothesis is then used to construct a method for its examination and also a computer program to analyse the collected data. I then undertake experimental work to validate my hypothesis, and compare my results and methods with a pre-established technique for interpreting the data, that of multi-dimensional scaling. Finally the method now shown to be feasible is discussed to indicate the its success and highlight its shortcomings. Further suggestions are also made as to further research avenues.