Two Tales of Persona in LLMs:
A Survey of Role-Playing and Personalization

Introduction

This is the official repository of the paper "Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization".

The concept of persona, originally adopted in dialogue literature, has re-surged as a promising framework for tailoring large language models (LLMs) to specific context (e.g., personalized search, LLM-as-a-judge). However, the growing research on leveraging persona in LLMs is relatively disorganized and lacks a systematic taxonomy. To close the gap, we present a comprehensive survey to categorize the current state of the field. We identify two lines of research, namely (1) LLM Role-Playing, where personas are assigned to LLMs, and (2) LLM Personalization, where LLMs take care of user personas. Additionally, we introduce existing methods for LLM personality evaluation. To the best of our knowledge, we present the first survey for role-playing and personalization in LLMs under the unified view of persona.

We continuously maintain this paper collection to foster future endeavors.

News

[2024.06.27] 🔥 We update an 8-page version on arXiv.
[2024.06.04] 🚀 Our paper is now available on arXiv and the reading list on GitHub.

🙆‍♀️ LLM Role-Play (Adapt to Environment)

LLMs are tasked to play the assigned personas (i.e., roles) and act accordance to environmental feedback.

The key aspect is how LLMs adapt to defined environments.

💼 Workshops

Date	Workshop	Website Link
2405	LLMAgent @ ICLR	ICLR 2024 Workshop on Large Language Model (LLM) Agents
2405	Agent Workshop @ CMU	CMU Agent Workshop 2024

🌎 Environments

💻 Software Development

Date	Authors	Venue	Paper
2308	Hong et al.	ICLR	MetaGPT: Meta Programming for Multi-Agent Collaborative Framework
2307	Qian et al.	arXiv	Communicative agents for software development
2305	Dong et al.	TOSEM	Self-collaboration code generation via chatgpt
2107	Chen et al.	arXiv	Evaluating large language models trained on code

🌐 Web

Date	Authors	Venue	Paper
2404	Liu et al.	arXiv	VisualWebBench: How Far Have Multimodal LLMs Evolved in Web Page Understanding and Grounding?
2401	Zheng et al.	LLMAgent @ ICLR	GPT-4V(ision) is a Generalist Web Agent, if Grounded
2401	Koh et al.	LLMAgent @ ICLR	VisualWebArena: Evaluating Multimodal Agents on Realistic Visual Web Tasks
2401	Cheng et al.	LLMAgent @ ICLR	SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents
2312	Gur et al.	EMNLP	Understanding HTML with Large Language Models
2312	Hong et al.	arXiv	CogAgent: A Visual Language Model for GUI Agents
2307	Gur et al.	ICLR	A Real-World WebAgent with Planning, Long Context Understanding, and Program Synthesis
2307	Zhou et al.	ICLR	WebArena: A Realistic Web Environment for Building Autonomous Agents
2306	Deng et al.	NeurIPS	Mind2web: Towards a generalist agent for the web
2303	Kim et al.	NeurIPS	Language Models can Solve Computer Tasks

🎮 Game

Date	Authors	Venue	Paper
2310	Wang et al.	EMNLP	Humanoid Agents: Platform for Simulating Human-like Generative Agents
2305	Wang et al.	TMLR	Voyager: An open-ended embodied agent with large language models
2305	Fu et al.	arXiv	Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedback
2304	Park et al.	UIST	Generative agents: Interactive simulacra of human behavior

🏥 Medical Application

Date	Authors	Venue	Paper
2312	Kwon et al.	AAAI	Large Language Models are Clinical Reasoners: Reasoning-Aware Diagnosis Framework with Prompt-Generated Rationales
2311	Tang et al.	arXiv	Medagents: Large language models as collaborators for zero-shot medical reasoning
2307	Wu et al.	ICLR	Large Language Models Perform Diagnostic Reasoning
2207	Liévin et al.	arXiv	Can large language models reason about medical questions?

🧑‍⚖️ LLM as Evaluators

Date	Authors	Venue	Paper
2308	Chan et al.	ICLR	ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate
2303	Wu et al.	NLPCC	Large Language Models are Diverse Role-Players for Summarization Evaluation

📦 General Framework

Date	Authors	Venue	Paper
2405	Ahn, et al	ACL Findings	TimeChara: Evaluating Point-in-Time Character Hallucination of Role-Playing Large Language Models
2308	Chen et al.	ICLR	AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
2307	Wang et al.	NAACL	Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration
2303	Li et al.	NeurIPS	CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society

🤖 Interaction & Behaviors

📊 Schemas

👤 Single-Agent

Date	Authors	Venue	Paper
2401	Cheng et al.	LLMAgent @ ICLR	SeeClick: Harnessing GUI Grounding for Advanced Visual GUI Agents
2401	Zheng et al.	LLMAgent @ ICLR	GPT-4V(ision) is a Generalist Web Agent, if Grounded
2312	Hong et al.	arXiv	CogAgent: A Visual Language Model for GUI Agents
2305	Wang et al.	TMLR	Voyager: An open-ended embodied agent with large language models

👥 Multi-Agent

Date	Authors	Venue	Paper
2311	Tang et al.	arXiv	Medagents: Large language models as collaborators for zero-shot medical reasoning
2308	Chen et al.	ICLR	AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
2308	Hong et al.	ICLR	MetaGPT: Meta Programming for Multi-Agent Collaborative Framework
2308	Chan et al.	ICLR	ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debate
2307	Qian et al.	arXiv	Communicative agents for software development
2305	Fu et al.	arXiv	Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedback

💡 Emergent Behaviors

Date	Authors	Venue	Paper
2311	Tang et al.	arXiv	Medagents: Large language models as collaborators for zero-shot medical reasoning
2308	Chen et al.	ICLR	AgentVerse: Facilitating Multi-Agent Collaboration and Exploring Emergent Behaviors
2307	Wang et al.	NAACL	Unleashing the Emergent Cognitive Synergy in Large Language Models: A Task-Solving Agent through Multi-Persona Self-Collaboration
2305	Fu et al.	arXiv	Improving Language Model Negotiation with Self-Play and In-Context Learning from AI Feedback
2303	Li et al.	NeurIPS	CAMEL: Communicative Agents for "Mind" Exploration of Large Language Model Society

🙆‍♂️ LLM Personalization (Adapt to User)

LLMs are tasked to take care of users’ personas (e.g., background information, or historical behaviors) to meet customized needs.

The key aspect is how LLMs adapt to distinct users.

💼 Workshops & Competitions

Date	Authors	Venue	Paper
2403	Deshpande et al.	PERSONALIZE @ EACL	Proceedings of the 1st Workshop on Personalization of Generative AI Systems (PERSONALIZE 2024)
2310	Chen et al.	Personalized Generative AI @ CIKM	The First Workshop on Personalized Generative AI @ CIKM 2023: Personalization Meets Large Language Models
1902	Dinan et al.	ConvAI2 @ NeurIPS	The Second Conversational Intelligence Challenge (ConvAI2)
1808	Yusupov et al.	ConvAI @ NeurIPS	NIPS Conversational Intelligence Challenge 2017 Winner System: Skill-based Conversational Agent with Supervised Dialog Manager

📌 Tasks

💬 Personalized Dialogue

🔧 ToD Modeling

LLMs Era

Date	Authors	Venue	Paper
2305	Yang et al.	EMNLP	RefGPT: Dialogue Generation of GPT, by GPT, and for GPT
2302	Li et al.	NeurIPS	Guiding large language models via directional stimulus prompting
2005	Hosseini-Asl et al.	NeurIPS	A Simple Language Model for Task-Oriented Dialogue

Comprehensive Paper List

Date	Authors	Venue	Paper
2312	Xu et al.	EMNLP	Baize: An Open-Source Chat Model with Parameter-Efficient Tuning on Self-Chat Data
2309	Hu et al.	arXiv	Enhancing Large Language Model Induced Task-Oriented Dialogue Systems Through Look-Forward Motivated Goals
2308	Wu et al.	SIGDIAL	DiactTOD: Learning Generalizable Latent Dialogue Acts for Controllable Task-Oriented Dialogue Systems
2305	Yang et al.	EMNLP	RefGPT: Dialogue Generation of GPT, by GPT, and for GPT
2305	Bang et al.	ACL	Task-Optimized Adapters for an End-to-End Task-Oriented Dialogue System
2304	Ashby et al.	CHI	Personalized Quest and Dialogue Generation in Role-Playing Games: A Knowledge Graph and Language Model-based Approach
2304	Hudevcek et al.	SIGDIAL	Are Large Language Models All You Need for Task-Oriented Dialogue?
2302	Li et al.	NeurIPS	Guiding large language models via directional stimulus prompting
2302	Feng et al.	ICLR	Fantastic rewards and how to tame them: A case study on reward learning for task-oriented dialogue systems
2210	Huryn et al.	COLING	Automatic Generation of Large-scale Multi-turn Dialogues from Reddit
2108	Peng et al.	TACL	Soloist: Building Task Bots at Scale with Transfer Learning and Machine Teaching
2012	Yang et al.	AAAI	UBAR: Towards Fully End-to-End Task-Oriented Dialog Systems with GPT-2
2008	Madotto et al.	arXiv	Language models as few-shot learner for task-oriented dialogue systems
2005	Hosseini-Asl et al.	NeurIPS	A Simple Language Model for Task-Oriented Dialogue

Pre-LLMs Era

Date	Authors	Venue	Paper
2006	Jianhong Wang et al.	ICLR	Modelling hierarchical structure between dialogue policy and natural language generator with option framework for task-oriented dialogue system
1606	N. Mrksic et al.	ACL	Neural Belief Tracker: Data-Driven Dialogue State Tracking
1506	Alessandro Sordoni et al.	NAACL	A Neural Network Approach to Context-Sensitive Generation of Conversational Responses

Comprehensive Paper List

Date	Authors	Venue	Paper
2105	Sun et al.	SIGIR	Simulating user satisfaction for the evaluation of task-oriented dialogue systems
2006	Wang et al.	ICLR	Modelling hierarchical structure between dialogue policy and natural language generator with option framework for task-oriented dialogue system
2003	Yang et al.	IEEE	Multitask Learning and Reinforcement Learning for Personalized Dialog Generation: An Empirical Study
1912	Huang	AAAI	MALA: Cross-Domain Dialogue Generation with Action Learning
1910	Zhang et al.	*SEM	Find or Classify? Dual Strategy for Slot-Value Predictions on Multi-Domain Dialog State Tracking
1905	Wu et al.	ACL	Transferable Multi-Domain State Generator for Task-Oriented Dialogue Systems
1807	Lei et al.	ACL	Sequicity: Simplifying Task-oriented Dialogue Systems with Single Sequence-to-Sequence Architectures
1804	Liu et al.	NAACL	Dialogue Learning with Human Teaching and Feedback in End-to-End Trainable Task-Oriented Dialogue Systems
1712	Rastogi et al.	IEEE	Scalable Multi-Domain Dialogue State Tracking
1709	Wu et al.	AAAI	StarSpace: Embed all the things!
1606	Miller et al.	EMNLP	Key-Value Memory Networks for Directly Reading Documents
1606	Mrksic et al.	ACL	Neural Belief Tracker: Data-Driven Dialogue State Tracking
1506	Sordoni et al.	NAACL	A Neural Network Approach to Context-Sensitive Generation of Conversational Responses

📝 User Persona Modeling

Date	Authors	Venue	Paper
2405	Han	ACL	PSYDIAL: Personality-based Synthetic Dialogue Generation Using Large Language Models
2307	Tang et al.	ACL	Enhancing Personalized Dialogue Generation with Contrastive Latent Variables: Combining Sparse and Dense Persona
1807	Zhang et al.	ACL	Personalizing Dialogue Agents: I have a dog, do you have pets too?

Comprehensive Paper List

Date	Authors	Venue	Paper
2405	Han	ACL	PSYDIAL: Personality-based Synthetic Dialogue Generation Using Large Language Models
2401	Lotfi et al.	IEEE	PersonalityChat: Conversation Distillation for Personalized Dialog Modeling with Facts and Traits
2401	Kim et al.	EACL	Commonsense-augmented Memory Construction and Management in Long-term Conversations via Context-aware Persona Refinement
2308	Tu et al.	arXiv	CharacterChat: Learning towards Conversational AI with Personalized Social Support
2307	Tang et al.	ACL	Enhancing Personalized Dialogue Generation with Contrastive Latent Variables: Combining Sparse and Dense Persona
2307	Ahn, et al	ACL	MPCHAT: Towards Multimodal Persona-Grounded Conversation
2011	Zhong et al.	EMNLP	Towards Persona-Based Empathetic Conversational Models
2007	Wu et al.	ACL	Guiding Variational Response Generator to Exploit Persona
2007	Liu et al.	ACL	You Impress Me: Dialogue Generation via Mutual Persona Perception
1911	Zheng et al.	AAAI	A Pre-Training Based Personalized Dialogue Generation Model with Persona-Sparse Data
1911	Song et al.	AAAI	Generating Persona Consistent Dialogues by Exploiting Natural Language Inference
1807	Zhang et al.	ACL	Personalizing Dialogue Agents: I have a dog, do you have pets too?

🛒 Recommendation System

Date	Authors	Venue	Paper
2305	Yang et al.	arXiv	PALR: Personalization Aware LLMs for Recommendation
2304	Wang et al.	arXiv	Zero-Shot Next-Item Recommendation using Large Pretrained Language Models
2108	Li et al.	ACL	Personalized Transformer for Explainable Recommendation

Comprehensive Paper List

Date	Authors	Venue	Paper
2405	Hu et al.	WWW	Enhancing sequential recommendation via llm-based semantic embedding learning
2311	Chen et al.	arXiv	A Survey on Large Language Models for Personalized and Explainable Recommendations
2308	Wang et al.	arXiv	RecMind: Large Language Model Powered Agent For Recommendation
2308	Chu et al.	arXiv	Leveraging Large Language Models for Pre-trained Recommender Systems
2306	Li et al.	arXiv	Prompt Tuning Large Language Models on Personalized Aspect Extraction for Recommendations
2305	Yang et al.	arXiv	PALR: Personalization Aware LLMs for Recommendation
2305	Zhang et al.	arXiv	Recommendation as Instruction Following: A Large Language Model Empowered Recommendation Approach
2304	Wang et al.	arXiv	Zero-Shot Next-Item Recommendation using Large Pretrained Language Models
2208	Chen et al.	KDD	Personalized Chit-Chat Generation for Recommendation Using External Chat Corpora
2202	Li et al.	TOIS	Personalized Prompt Learning for Explainable Recommendation
2108	Li et al.	ACL	Personalized Transformer for Explainable Recommendation

🔍 Personalized Search

Date	Authors	Venue	Paper
2405	Zhou et al.	WWW	Cognitive personalized search integrating large language models with an efficient memory mechanism
2405	Baek et al.	WWW	Knowledge-Augmented Large Language Models for Personalized Contextual Query Suggestion
2405	Salemi	arXiv	Unified ranking for multiple retrieval-augmented large language models
2402	Sharma et al.	CHI	Generative echo chamber? effects of llm-powered search systems on diverse information seeking
2307	Eleni et al.	arXiv	Comparing Traditional and LLM-based Search for Consumer Choice: A Randomized Experiment
2307	Ziems et al.	ACL	Large Language Models are Built-in Autoregressive Search Engines
2107	Zhou et al.	SIGIR	Group based Personalized Search by Integrating Search Behaviour and Friend Network

🩺 Personalized Healthcare

Date	Authors	Venue	Paper
2402	Abbasian et al.	arXiv	Knowledge-Infused LLM-Powered Conversational Health Agent: A Case Study for Diabetes Patients
2402	Jin et al.	arXiv	Health-LLM: Personalized Retrieval-Augmented Disease Prediction System
2310	Abbasian et al.	arXiv	Conversational Health Agents: A Personalized LLM-Powered Agent Framework
2309	Zhang et al.	arXiv	LLM-based Medical Assistant Personalization with Short- and Long-Term Memory Coordination

📚 Personalized Education

Date	Authors	Venue	Paper
2403	Park et al.	CHI	Empowering personalized learning through a conversation-based tutoring system with student modeling
2308	Dan et al.	arXiv	Educhat: A large-scale language model-based chatbot system for intelligent education
2307	Shehata et al.	BEA @ ACL	Enhancing Video-based Learning Using Knowledge Tracing: Personalizing Students’ Learning Experience with ORBITS

🛠️ Methods

🎛️ Fine-Tuning

Date	Authors	Venue	Paper
2403	Mondal et al.	EACL	Presentations by the Humans and For the Humans: Harnessing LLMs for Generating Persona-Aware Slides from Documents
2402	Li et al.	arXiv	Personalized Language Modeling from Personalized Human Feedback
2402	Tan et al.	arXiv	Democratizing Large Language Models via Personalized Parameter-Efficient Fine-tuning
2312	Hwang et al.	arXiv	Promptable Behaviors: Personalizing Multi-Objective Rewards from Human Preferences
2312	Shea et al.	EMNLP	Building Persona Consistent Dialogue Agents with Offline Reinforcement Learning
2311	Qin et al.	arXiv	Enabling on-device large language model personalization with self-supervised data selection and synthesis
2310	Jang et al.	arXiv	Personalized large language model alignment via post-hoc parameter merging
2303	Kirk et al.	arXiv	Personalisation within bounds: A risk taxonomy and policy framework for the alignment of large language models with personalised feedback

🔗 Retrieval Augmentation

Date	Authors	Venue	Paper
2404	Zhang et al.	arXiv	Personalized LLM Response Generation with Parameterized Memory Injection
2403	Zhong et al.	AAAI	MemoryBank: Enhancing Large Language Models with Long-Term Memory
2402	Sun et al.	arXiv	Persona-DB: Efficient Large Language Model Personalization for Response Prediction with Collaborative Data Refinement
2402	Tan et al.	arXiv	Democratizing Large Language Models via Personalized Parameter-Efficient Fine-tuning
2205	Fu et al.	ACL	There Are a Thousand Hamlets in a Thousand People’s Eyes: Enhancing Knowledge-grounded Dialogue with Personal Memory
2106	Wu et al.	NAACL	Personalized Response Generation via Generative Split Memory Network

✍️ Prompting

📄 Vanilla Personalized Prompt

Date	Authors	Venue	Paper
2305	Dai et al.	RecSys	Uncovering ChatGPT’s Capabilities in Recommender Systems
2305	Christakopoulou et al.	arXiv	Large Language Models for User Interest Journeys
2305	Zhiyuli et al.	arXiv	BookGPT: A General Framework for Book Recommendation Empowered by Large Language Model

🔦 Retrieval-Augmented Personalized Prompt

Date	Authors	Venue	Paper
2311	Mysore et al.	arXiv	PEARL: Personalizing Large Language Model Writing Assistants with Generation-Calibrated Retrievers
2308	Li et al.	arXiv	Teach LLMs to Personalize -- An Approach inspired by Writing Education
2304	Salemi et al.	arXiv	LaMP: When Large Language Models Meet Personalization

📂 Profile-Augmented Prompt

Date	Authors	Venue	Paper
2405	Li et al.	WWW	Learning to Rewrite Prompts for Personalized Text Generation
2310	Richardson et al.	arXiv	Integrating Summarization and Retrieval for Enhanced Personalization via Large Language Models
2305	Liu et al.	WSDM	ONCE: Boosting Content-based Recommendation with Both Open- and Closed-source Large Language Models

🧐 LLM Personality Evaluation

Date	Authors	Venue	Paper
2401	Huang et al.	ICLR	On the Humanity of Conversational AI: Evaluating the Psychological Portrayal of LLMs
2309	Jiang et al.	NeurIPS	Evaluating and inducing personality in pre-trained language models
2307	Fang et al.	ACL	On Text-based Personality Computing: Challenges and Future Directions

Comprehensive Paper List

Date	Authors	Venue	Paper
2403	Sorokovikova et al.	PERSONALIZE @ EACL	LLMs Simulate Big5 Personality Traits: Further Evidence
2403	Frisch et al.	PERSONALIZE @ EACL	LLM Agents in Interaction: Measuring Personality Consistency and Linguistic Alignment in Interacting Populations of Large Language Models
2402	Song et al.	arXiv	Identifying Multiple Personalities in Large Language Models with External Evaluation
2402	Song et al.	arXiv	Identifying Multiple Personalities in Large Language Models with External Evaluation
2402	Yang et al.	arXiv	LLM Agents for Psychology: A Study on Gamified Assessments
2401	Huang et al.	ICLR	On the Humanity of Conversational AI: Evaluating the Psychological Portrayal of LLMs
2312	Rao et al.	EMNLP	Can ChatGPT Assess Human Personalities? A General Evaluation Framework
2311	Dorner et al.	SoLaR @ NeurIPS	Do personality tests generalize to large language models?
2310	Wang et al.	arXiv	InCharacter: Evaluating Personality Fidelity in Role-Playing Agents through Psychological Interviews
2309	Jiang et al.	NeurIPS	Evaluating and inducing personality in pre-trained language models
2307	Pan et al.	arXiv	Do LLMs Possess a Personality? Making the MBTI Test an Amazing Evaluation for Large Language Models
2307	Fang et al.	ACL	On Text-based Personality Computing: Challenges and Future Directions
2307	Ji et al.	arXiv	Is ChatGPT a Good Personality Recognizer? A Preliminary Study
2305	Jiang et al.	arXiv	Personallm: Investigating the ability of large language models to express big five personality traits

🌱 How to contribute

✨ Welcome to contribute to this reading list via 📝 Issues using the following format.

Date	Authors	Venue	Paper
1706	Vaswani, et al	NeurIPS	Attention Is All You Need

🔖 Citation

📚 If you find our survey beneficial for your research, please kindly cite our paper :-)

@misc{tseng2024talespersonallmssurvey,
  title={Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization},
  author={Yu-Min Tseng and Yu-Chao Huang and Teng-Yun Hsiao and Wei-Lin Chen and Chao-Wei Huang and Yu Meng and Yun-Nung Chen},
  year={2024},
  eprint={2406.01171},
  archivePrefix={arXiv},
  primaryClass={cs.CL},
  url={https://arxiv.org/abs/2406.01171},
}

🖌️ Authors

Yu-Min Tseng*, Yu-Chao Huang*, Teng-Yun Hsiao*, Wei-Lin Chen*, Chao-Wei Huang, Yu Meng, Yun-Nung Chen.

(* Equal Contribution.) (Acknowlegement: Yu-Ching Hsu, Jia-Yin Foo.)

Two Tales of Persona in LLMs: A Survey of Role-Playing and Personalization

Introduction

News

Table of Contents

🙆‍♀️ LLM Role-Play (Adapt to Environment)

💼 Workshops

🌎 Environments

💻 Software Development

🌐 Web

🎮 Game

🏥 Medical Application

🧑‍⚖️ LLM as Evaluators

📦 General Framework

🤖 Interaction & Behaviors

📊 Schemas

👤 Single-Agent

👥 Multi-Agent

💡 Emergent Behaviors

🙆‍♂️ LLM Personalization (Adapt to User)

💼 Workshops & Competitions

📌 Tasks

💬 Personalized Dialogue

🔧 ToD Modeling

📝 User Persona Modeling

🛒 Recommendation System

🔍 Personalized Search

🩺 Personalized Healthcare

📚 Personalized Education

🛠️ Methods

🎛️ Fine-Tuning

🔗 Retrieval Augmentation

✍️ Prompting

📄 Vanilla Personalized Prompt

🔦 Retrieval-Augmented Personalized Prompt

📂 Profile-Augmented Prompt

🧐 LLM Personality Evaluation

🌱 How to contribute

🔖 Citation

🖌️ Authors

About

Two Tales of Persona in LLMs:
A Survey of Role-Playing and Personalization