Quantum projections on conceptual subspaces: A deeper dive into methodological challenges and opportunities.

Area of Science:

• Cognitive Science
• Computational Linguistics
• Quantum Cognition

Background:

• The distributional hypothesis of language suggests meaning is derived from context.
• Previous work established methods for generating conceptual subspaces from textual data.
• Quantum probability models offer explanations for psychological biases in similarity judgments, like violations of symmetry and triangular inequality.

Purpose of the Study:

• To clarify methodological aspects of using quantum projections on conceptual subspaces for similarity research.
• To discuss theoretical and practical implications of this data-driven approach.
• To propose novel methods for generating subspaces and present new empirical evidence on similarity assumption violations.

Main Methods:

• Generating conceptual subspaces from textual information using a data-driven methodology.
• Utilizing the quantum model of similarity to analyze these subspaces.
• Proposing the use of conceptual or contextual contours to define subspaces, leading to Aggregated Terms Subspaces (ATSs), Aggregated Contexts Subspaces (ACSs), and Aggregated Features Subspaces (AFSs).

Main Results:

• Demonstrated the empirical examination of violations of symmetry and triangular inequality within a data-driven framework.
• Provided new data confirming the violation of the triangular inequality assumption.
• Applied the quantum similarity model to Aggregated Terms Subspaces (ATSs) to illustrate these violations.

Conclusions:

• The quantum projection methodology on conceptual subspaces provides a powerful tool for studying psychological biases in similarity.
• The findings highlight limitations of classical geometric models for psychological similarity.
• Future research can leverage different subspace generation techniques (ATSs, ACSs, AFSs) to further explore context and feature effects in similarity judgments.