Improving the Predictive Capability of Type-Token Models

This project will assess the predictive capability of type-token models using a range of complex systems, with a view to identifying their respective strengths and weaknesses. The initial sources of data are expected to be textual documents, but other type-token systems (natural and artificial) will be identified and their behaviours quantitatively compared. It is intended that a sufficient body of data will yield hitherto unknown insights, which may be incorporated into improved models for greater predictive accuracy. It is also hoped that new applications for these models will also be identified and assessed.