A Knowledge Graph and Large Language Model Approach for Mapping Associations Between Cognitive Neuroscience and Marketing.

A semi-automated framework to create a structured association map between cognitive neuroscience and marketing, using expert-validated large language models (LLMs), supports a knowledge graph-based approach. The starting point for this framework is an existing cognitive neuroscience knowledge graph (CNKG) of ۳۱۲ concepts as the source domain and a curated marketing concept list consisting of ۱۰۰ concepts as the target domain. The mapping process consists of two distinct phases: the first phase involves automated association generation via four different LLMs (GPT-۴, Gemini, Claude, & Grok) producing candidate associations between cognitive neuroscience and marketing by determining synonymous relationships with their word choices; and the second phase includes human-based expert validation in which marketing and economics experts evaluate, revise, and filter the candidate associations produced by the automated associations. This further emphasizes that human supervision is essential in maintaining the scientific integrity of the mapping process. In reviewing and analyzing the graph that was produced through this process, many core concepts within neuroscience were identified as being hub nodes (reward processing, emotion, attention, and decision-making). Additionally, there were strong relationships between the hub neuroscience concepts and many marketing constructs (including the brand).

کلیدواژه ها:

knowledge graph ، Large Language Models ، Cognitive neuroscience ، Neuromarketing ، Cross-domain Knowledge Mapping