“Integrating LLMs and RAG for Improved Usability in Multimedia Archives” presented by Carroll
Symposium:
Session Title:
- Summit on New Media Art Archiving (short papers)
Presentation Title:
- Integrating LLMs and RAG for Improved Usability in Multimedia Archives
Presentation Subtheme:
- Fourth Summit on New Media Art Archiving
Presenter(s):
Venue(s):
Abstract:
This paper explores the roles of generative AI technologies, such as Large Language Models (LLMs) and Retrieval Augmentation Generation (RAG) systems to enhance multimedia archives. In it, archives are redefined as dynamic entities for knowledge construction and validation, focusing on improving cultural accessibility and interaction. The paper addresses the challenges in archiving new media art, emphasising how semantic embeddings and LLMs make user interactions with archives more intuitive and accessible. The practical implementation of these technologies is explored in the Computer Art Archive’s pilot program, which incorporates this technology into systems akin to current knowledge retrieval frameworks. The paper concludes by discussing how these advancements democratise knowledge access and contribute to the enrichment of digital cultural heritage.
