Chapter 9 - Conceptual spaces and scientific data models

Abstract

This chapter will more substantially develop our approach to Conceptual Space Theory in natural (as well as programming) language contexts, that was initiated in earlier chapters. We present further philosophical motivations for the structural details of our proposed “Syntagmatic Graph” representations and examine techniques for integrating conceptual spaces with linguistic paradigms such as Conceptual Role Semantics and situational semantics. Our central argument is that the classical linguistic concept of “thematic roles” provides an alternative framework for analyzing the semantic integration of multiple conceptual spaces, contrasted with “quantitative blend” models endemic to conceptual space theory proper. Therefore we propose “role-indexed” Conceptual Space models, which have distinct semantic and syntactic patterns, juxtaposing this theory to existing formalizations of conceptual spaces in (for example) Quantum NLP. With that natural-language foundation as a motivation, we then consider semantic models as they could be more concretely applied to scientific data sets.