Large Language Models as Recommendation Systems in Museums

Abstract

This paper proposes the utilization of large language models as recommendation systems for museum visitors. Since the aforementioned models lack the notion of context, they cannot work with temporal information that is often present in recommendations for cultural environments (e.g., special exhibitions or events). In this respect, the current work aims to enhance the capabilities of large language models through a fine-tuning process that incorporates contextual information and user instructions. The resulting models are expected to be capable of providing personalized recommendations that are aligned with user preferences and desires. More specifically, Generative Pre-trained Transformer 4, a knowledge-based large language model is fine-tuned and turned into a context-aware recommendation system, adapting its suggestions based on user input and specific contextual factors such as location, time of visit, and other relevant parameters. The effectiveness of the proposed approach is evaluated through certain user studies, which ensure an improved user experience and engagement within the museum environment.

Will Artificial Intelligence Affect How Cultural Heritage Will Be Managed in the Future? Responses Generated by Four genAI Models

“…Digitised physical heritage assets (sites, buildings, and objects) can be utilised within a genAI environment while preserving provenance, authenticity, and veracity [63]. In the museum space, genAI language models can be successfully employed in the creation of exhibit labels and catalogue information [64], developing visitor query systems and guides for museums [65][66][67], conceptualising entire exhibitions [68], as well as exploring how to approach aspects such as memorialisation [69].…”

A Holistic Approach for Enhancing Museum Performance and Visitor Experience

“…The proposed ontologies developed in the context of MELTOPENLAB rely on specific ontology engineering principles, defining the purpose of the ontology, and how it contributes to expanding cultural knowledge for museum staff and visitors [44,45]. In the MELTOPENLAB case, ontologies do not work as standalone and static repositories of knowledge; instead, their scope within the AR is to exchange information in a structured and manageable manner with the other MELTOPENLAB components.…”

LLaMALoop: Enhancing Information Retrieval in LLaMA with Semantic Relevance Feedback Loop

“…The evolution of LLMs has been remarkable, starting from initial models that laid the groundwork for natural language processing to recent sophisticated systems capable of intricate language understanding [12,10,13]. The early stages of LLM development focused on foundational algorithms where models like RNNs and LSTMs showed potential in sequential data processing [13,14,15,16]. This was a significant leap, demonstrating the ability of machines to grasp basic language structures.…”

“…The application of LLMs in domain-specific contexts also garnered interest. Models were fine-tuned and evaluated in specialized fields like legal document analysis, medical diagnosis from patient records, and even creative writing, showing their versatility [13,25,26,27,28]. The role of unsupervised and semi-supervised learning in LLMs was another area of focus, reducing the dependence on labeled data and exploring more efficient ways of training [29,17,30].…”