The paper introduces a Situation–Internal state–Behavior Persona method to improve large language models’ ability to impersonate real individuals in social media contexts. It also proposes an evaluation protocol that supplies LLM evaluators with reference information about the target individual. Experiments on a new dataset of social media replies show the method surpasses state‑of‑the‑art in‑context learning baselines, and the protocol correlates moderately with human judgments, while additional tests on fictional characters confirm broader applicability.
arXiv:2603. 03915v2 Announce Type: replace-cross Abstract: Large Language Models (LLMs) have shown remarkable potential in developing role-playing agents (RPAs).
Personalizing large language models (LLMs) to individual users is essential for improving user experience, yet existing approaches typically rely on explicit preference supervision such as pairwise comparisons or demographic attributes, limiting their applicability in natural interaction settings. We propose IRIS, a framework that learns dynamic user personas directly from implicit interaction streams by extracting behavioral signals from everyday conversations and iteratively refining persona representations through a prediction-driven closed loop without requiring explicit feedback.
arXiv:2607. 26473v1 Announce Type: new Abstract: Personalizing large language models (LLMs) to individual users is essential for improving user experience, yet existing approaches typically rely on explicit preference supervision such as pairwise comparisons or demographic attributes, limiting their applicability in natural interaction settings.
arXiv:2609.22255v1 Announce Type: new Abstract: Existing approaches to persona simulation with Large Language Models (LLMs) mostly rely on shallow character descriptions that fail to sustain coherent...
arXiv:2610.08513v1 Announce Type: cross Abstract: LLMs are increasingly deployed as autonomous agents in social environments, making it critical to study their ability to faithfully simulate human in...
The paper examines whether fine‑tuning large language models (LLMs) with personality‑labelled data improves their ability to act as socially interactive agents. Two small open‑weight LLMs were fine‑tuned on a corpus of personality‑labelled social media posts and dialogues, and the resulting models were evaluated in various social interaction scenarios by independent LLM judges. The findings show that the fine‑tuned models do not outperform their baseline counterparts in role‑playing personalities, though they offer comparable text quality and increased linguistic diversity for the Qwen models; low inter‑rater agreement limits confidence in the results, suggesting future work should focus on training data quality and domain alignment.
The paper introduces PRISM, a new framework for evaluating how well large language models (LLMs) maintain persona fidelity in dynamic dialogue. PRISM reframes the task as a structured inverse inference problem grounded in Systemic Functional Linguistics, breaking persona fidelity into Task Framing, Interpersonal Stance, and Linguistic Style dimensions. Experiments demonstrate that PRISM produces more accurate and stable judgments than existing holistic or static psychometric methods, offering a more reliable and auditable evaluation process.
Personalized language-model assistants are often evaluated through a memory lens: can a model recall preferences users have explicitly stated in dialogue? More comprehensive personalization demands a harder capability -- inferring what users care about from the multimodal traces they naturally leave behind.
arXiv:2606. 02754v1 Announce Type: new Abstract: Personalization is a crucial capability of modern language agents.
arXiv:2605. 09159v2 Announce Type: replace Abstract: Recent work shows that large language models (LLMs) encode behavioral traits ("personas") as linear directions in activation space, often called "persona vectors".
arXiv:2610.20548v1 Announce Type: new Abstract: Understanding the social intelligence of large language models (LLMs) is increasingly important for applications involving human-AI interaction and mul...
The paper introduces ACTOR, a framework that automatically constructs action trajectories for literary role‑playing agents by extracting dialogue and non‑dialogue behaviors from novels and representing them with a Thought‑Action‑Observation structure. It builds a dataset from 143 literary works, generates counterfactual behaviors, and trains models using Regularized Preference Optimization to improve character‑consistent action prediction. Evaluation on 17,560 instances shows that the ACTOR‑trained model significantly outperforms baselines in next‑behavior selection, demonstrating the value of temporally organized supervision for literary role‑playing.