LLM-Based Storytelling Agent for Older Adults Integrates Argument Mining
A research paper on arXiv (2605.10531) presents a reflective storytelling agent for older adults that combines knowledge graphs, user modeling, argumentation theory, and argument mining to guide LLM-based narrative generation. The system aims to address LLM limitations such as hallucinations and lack of transparency. Phase I involved participatory design with 11 domain experts for formative evaluation. Phase II tested the system with 55 older adults across four prompts and two creativity levels, assessing purpose, usefulness, cultural relatability, and consistency. The agent generates narratives grounded in structured user models of health-promoting activities and motivations.
Key facts
- arXiv paper 2605.10531
- Reflective storytelling agent for older adults
- Integrates knowledge graphs, user modeling, argumentation theory, argument mining
- Addresses LLM hallucinations and limited transparency
- Phase I: participatory design with 11 domain experts
- Phase II: evaluation with 55 older adults
- Four prompts and two creativity levels tested
- Narratives based on health-promoting activities and motivations
Entities
Institutions
- arXiv