# Controlling and Assessing Appropriate Persona Use in LLM-based Dialogue Generation

arXiv: https://arxiv.org/abs/2609.04676 · HTML: https://mlsanity.com/paper/2609.04676/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 31 of its 44 references that are in the map.
- First posted on arXiv 4 Sep 2026.

## Builds on (most cited first, 30 of 31)

- [BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding](https://mlsanity.com/paper/1810.04805/) (2018)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [BERTScore: Evaluating Text Generation with BERT](https://mlsanity.com/paper/1904.09675/) (2019)
- [Qwen2.5 Technical Report](https://mlsanity.com/paper/2412.15115/) (2024)
- [Large Language Models are Zero-Shot Reasoners](https://mlsanity.com/paper/2205.11916/) (2022)
- [Self-Refine: Iterative Refinement with Self-Feedback](https://mlsanity.com/paper/2303.17651/) (2023)
- [Steering Llama 2 via Contrastive Activation Addition](https://mlsanity.com/paper/2312.06681/) (2023)
- [Personalizing Dialogue Agents: I have a dog, do you have pets too?](https://mlsanity.com/paper/1801.07243/) (2018)
- [Decomposed Prompting: A Modular Approach for Solving Complex Tasks](https://mlsanity.com/paper/2210.02406/) (2022)
- [MoverScore: Text Generation Evaluating with Contextualized Embeddings and Earth Mover Distance](https://mlsanity.com/paper/1909.02622/) (2019)
- [A Persona-Based Neural Conversation Model](https://mlsanity.com/paper/1603.06155/) (2016)
- [Dialogue Natural Language Inference](https://mlsanity.com/paper/1811.00671/) (2018)
- [Don't Say That! Making Inconsistent Dialogue Unlikely with Unlikelihood Training](https://mlsanity.com/paper/1911.03860/) (2019)
- [Personalizing Dialogue Agents via Meta-Learning](https://mlsanity.com/paper/1905.10033/) (2019)
- [Learning to Select Knowledge for Response Generation in Dialog Systems](https://mlsanity.com/paper/1902.04911/) (2019)
- [BoB: BERT Over BERT for Training Persona-based Dialogue Models from Limited Personalized Data](https://mlsanity.com/paper/2106.06169/) (2021)
- [Few-Shot Bot: Prompt-Based Learning for Dialogue Systems](https://mlsanity.com/paper/2110.08118/) (2021)
- [Exploiting Persona Information for Diverse Generation of Conversational Responses](https://mlsanity.com/paper/1905.12188/) (2019)
- [Faithful Persona-based Conversational Dataset Generation with Large Language Models](https://mlsanity.com/paper/2312.10007/) (2023)
- [Generating Persona Consistent Dialogues by Exploiting Natural Language Inference](https://mlsanity.com/paper/1911.05889/) (2019)
- [Personalized Text Generation with Contrastive Activation Steering](https://mlsanity.com/paper/2503.05213/) (2025)
- [A Wind of Change: Detecting and Evaluating Lexical Semantic Change across Times and Domains](https://mlsanity.com/paper/1906.02979/) (2019)
- [Exploring Prompt-based Few-shot Learning for Grounded Dialog Generation](https://mlsanity.com/paper/2109.06513/) (2021)
- [A Personalized Dialogue Generator with Implicit User Persona Detection](https://mlsanity.com/paper/2204.07372/) (2022)
- [Building Persona Consistent Dialogue Agents with Offline Reinforcement Learning](https://mlsanity.com/paper/2310.10735/) (2023)
- [Dialogue Language Model with Large-Scale Persona Data Engineering](https://mlsanity.com/paper/2412.09034/) (2024)
- [Ontologically Faithful Generation of Non-Player Character Dialogues](https://mlsanity.com/paper/2212.10618/) (2022)
- [PATS: Personality-Aware Teaching Strategies with Large Language Model Tutors](https://mlsanity.com/paper/2601.08402/) (2026)

## Abstract (arXiv)

In persona-based dialogue generation (PDG), LLMs often overuse persona attributes by incorporating them regardless of dialogue context, resulting in unnatural responses. Despite its practical significance, the underlying causes remain unexplored, with no method to mitigate this problem or metric to assess the appropriateness of persona use. To address these issues, we first conduct a comprehensive analysis of LLM-based PDG, revealing that LLMs exhibit a systematic bias to incorporate all given persona attributes, and that existing metrics fail to capture contextual appropriateness. Building on these findings, we propose Self-CONtrastive Persona Overuse Suppression (SCONPOS) to mitigate overuse by directly intervening in LLMs' internal representations at the prompt encoding stage, without requiring any response generation. We further propose the Persona Appropriateness Score (PAS), a novel metric that penalizes both overuse and underuse. Experimental results demonstrate that SCONPOS systematically reduces overuse, and PAS captures the contextual appropriateness of persona use.