# A Reflective Storytelling Agent for Older Adults: Integrating Argumentation Schemes and Argument Mining in LLM-Based Personalised Narratives

arXiv: https://arxiv.org/abs/2605.10531 · HTML: https://mlsanity.com/paper/2605.10531/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Medical and scientific domain agents, one of 16 directions in LLM agents and tool use.
- Builds on 5 of its 67 references that are in the map.
- First posted on arXiv 11 May 2026.

## Placed in

- LLM agents and tool use / [Medical and scientific domain agents](https://mlsanity.com/agents/medical-and-scientific-domain-agents/)

## Builds on (most cited first, 5 of 5)

- [Neural-Symbolic Learning and Reasoning: A Survey and Interpretation](https://mlsanity.com/paper/1711.03902/) (2017)
- [Cue-CoT: Chain-of-thought Prompting for Responding to In-depth Dialogue Questions with LLMs](https://mlsanity.com/paper/2305.11792/) (2023)
- [On the Way to LLM Personalization: Learning to Remember User Conversations](https://mlsanity.com/paper/2411.13405/) (2024)
- [Large Language Models in Argument Mining: A Survey](https://mlsanity.com/paper/2506.16383/) (2025)
- [LLMs for Argument Mining: Detection, Extraction, and Relationship Classification of pre-defined Arguments in Online Comments](https://mlsanity.com/paper/2505.22956/) (2025)

## Abstract (arXiv)

This work investigates knowledge-driven, large language model (LLM)-based storytelling for older adults. We present a reflective storytelling agent integrating knowledge graphs, user modelling, and argumentation theory to guide person-tailored narratives. Argument mining supports inspection of argument structure, while persona grounding is assessed separately against structured information. The study comprised three phases. Phase I involved 11 domain experts whose feedback informed system refinement. Phase II involved 55 older adults evaluating narratives based on predefined personas across inquiry, deliberation, and persuasion purposes and two creativity levels. Phase III compared the full system with an argumentation-scheme ablation and standard LLM baseline, and validated zero-shot argument mining against human-consensus annotations. In Phase II, the intended purpose was recognised in 30.6% of 330 assessments; 33.3% indicated no clear purpose and 36.1% another purpose. Cultural relatability was descriptively associated with future-use intention, although 30 of 55 participants did not intend to use similar functionality. In Phase III, 67.6% of applicable sampled statements were persona-grounded in the full system versus 57.9% without scheme guidance, a non-significant difference. Scheme guidance increased explicit premises without changing claim counts. The full system showed substantially greater sampled grounding than the standard LLM baseline. Zero-shot argument mining showed moderate agreement with human-consensus annotations, supporting its role as an inspection aid rather than a verifier. These findings highlight the potential and limitations of reflective, person-tailored LLM storytelling for older adults.