Survey-Evolution-DS

This is the repo which records the evolution of LM-based dialogue system. We list works in each stage, and will constantly update it, welcome to raise a issue to add new works!!

• Task-oriented Dialogue System (TOD)
  • Natural Language Understanding (NLU)
  • Dialogue State Tracking (DST)
  • Dialogue Policy Learning (DST)
  • Natural Language Generation (NLG)
  • End-to-End TOD (E2E TOD)
• Open-domain Dialogue System (ODD)
• Unified Dialogue System (UniDS)

The Evolution of LM-based Dialogue System

Survey Paper

• A Survey of Language Model-based Dialogue System 🔥🔥🔥🔥🔥 we write a blog for better understanding: TOD ODD PLM LLM click here.

• End-to-end Task-oriented Dialogue: A Survey of Tasks, Methods, and Future Directions E2E TOD EMNLP 2023 🔥🔥🔥

• Recent advances in deep learning based dialogue systems: a systematic survey Artificial Intelligence Review 2023 🔥🔥🔥

• A Survey on Recent Advances and Challenges in Reinforcement Learning Methods for Task-oriented Dialogue Policy LearningDPL Machine Intelligence Research 2023 🔥

• A Survey on Proactive Dialogue Systems: Problems, Methods, and Prospects ODD IJCAI 2023 🔥🔥

• Let's Negotiate! A Survey of Negotiation Dialogue Systems ODD Arxiv 2022

• Recent advances and challenges in task-oriented dialog systemsTOD SCTC 2020

• Challenges in Building Intelligent Open-domain Dialog Systems ODD TOIS 2020

• A Survey on Dialogue Systems: Recent Advances and New Frontiers TOD ODD SIGKDD 2017

1st Stage -- SLM: Early Stage

• Eliza, Alice, GUS

2nd Stage -- NLM: Independent Development

• End-to-End Learning of Task-Oriented DialogsE2E TOD NAACL 2018 first E2E TOD

• Assigning Personality/Profile to a Chatting Machine for Coherent Conversation Generation ODD IJCAI 2018

3rd Stage -- PLM: Fusion Starts!

• Improving Factual Consistency for Knowledge-Grounded Dialogue Systems via Knowledge Enhancement and Alignment ODD EMNLP 2023 🔥🔥

• Re3Dial: Retrieve, Reorganize and Rescale Conversations for Long-Turn Open-Domain Dialogue Pre-training ODD EMNLP 2023 🔥🔥🔥

• DiSTRICT: Dialogue State Tracking with Retriever Driven In-Context Tuning DST EMNLP 2023 T5 model

• Well Begun is Half Done: Generator-agnostic Knowledge Pre-Selection for Knowledge-Grounded Dialogue ODD EMNLP 2023 BART, T5

• Turn-Level Active Learning for Dialogue State Tracking RL EMNLP 2023

• JoTR: A Joint Transformer and Reinforcement Learning Framework for Dialog Policy Learning DPL TOD Arxiv 2023

• Retrieval-free Knowledge Injection through Multi-Document Traversal for Dialogue Models ODD ACL 2023

• Learning to Generate Prompts for Dialogue Generation through Reinforcement Learning ODD Arixv 2022

• Integrating Pretrained Language Model for Dialogue Policy Evaluation DPL TOD ICASSP 2022 🔥🔥🔥 first work of RLAIF in DPL

• Personalized Dialogue Generation with Persona-Adaptive Attention ODD AAAI 2023

3.1: Fusions within TOD

• Modularized Pre-Training for End-to-End Task-Oriented Dialogue E2E TOD TASLP 2023

• PPTOD: Multi-Task Pre-Training for Plug-and-Play Task-Oriented Dialogue System E2E TOD ACL 2022

• Soloist: Building Task Bots at Scale with Transfer Learning and Machine Teaching E2E TOD TACL 2021

• MOSS: End-to-End Dialog System Framework with Modular Supervision AAAI 2020 first work for modular E2E TOD

3.2: Fusion between TOD with ODD

• Dual-Feedback Knowledge Retrieval for Task-Oriented Dialogue Systems EMNLP 2023 T5 model, related to Q-TOD

• Continual Dialogue State Tracking via Example-Guided Question Answering EMNLP 2023 T5 model

• Enabling Semi-Structured Knowledge Access via a Question-Answering Module in Task-oriented Dialogue Systems QA -> TOD CUI 2023

• Q-TOD: A Query-driven Task-oriented Dialogue System TOD -> ODD EMNLP 2022

• UniDS: A Unified Dialogue System for Chit-Chat and Task-oriented Dialogues ODD -> TOD DialDoc 2022

• GODEL: Large-Scale Pre-Training for Goal-Directed Dialog TOD -> ODD Arxiv 2022 [Code]

3.3: Fusion between DM and LLM

• LLaMA2-Chat Llama 2: Open Foundation and Fine-Tuned Chat Models Arxiv 2023

• Parrot: Enhancing Multi-Turn Chat Models by Learning to Ask Questions Arxiv 2023 multi-turn instruction-tuning data construction

• Enhancing Chat Language Models by Scaling High-quality Instructional Conversations EMNLP 2023

• BlenderBot 3: a deployed conversational agent that continually learns to responsibly engage Arxiv 2022

• Pangu-Bot: Efficient Generative Dialogue Pre-training from Pre-trained Langauge Model Arxiv 2022

Others

• Improving the Robustness of Knowledge-Grounded Dialogue via Contrastive Learning target noises in the input such as kg or misspelling in the query ODD

4nd Stage -- LLM-based Dialogue System

4.1: Internal Reasoning

• Investigating Content Planning for Navigating Trade-offs in Knowledge-Grounded Dialogue content planning similar with TPE

• [COOPER: Coordinating Specialized Agents towards a Complex Dialogue Goal] AAAI 2024 multi-agent cue-cot？

-Reverse Multi-Choice Dialogue Commonsense Inference with Graph-of-Thought AAAI 2024

• Cue-CoT: Chain-of-thought Prompting for Responding to In-depth Dialogue Questions with LLMs ODD EMNLP 2023 🔥🔥🔥 linguistic cues

• Symbolic Planning and Code Generation for Grounded Dialogue TOD EMNLP 2023 [code] interesting

• Scalable-DSC: A Structural Template Prompt Approach to Scalable Dialogue State Correction EMNLP 2023

• Mirages: On Anthropomorphism in Dialogue Systems ODD EMNLP 2023 linguistic cues

• Prompt-Based Monte-Carlo Tree Search for Goal-Oriented Dialogue Policy Planning TOD DPL EMNLP 2023

Proactive

• Prompting and Evaluating Large Language Models for Proactive Dialogues: Clarification, Target-guided, and Non-collaboration ODD EMNLP 2023

• Target-oriented Proactive Dialogue Systems with Personalization: Problem Formulation and Dataset Curation ODD EMNLP 2023

• Rethinking Conversational Agents in the Era of LLMs: Proactivity, Non-collaborativity, and Beyond ODD SIGIR-AP 2023

Empathetic Dialogue

• EmoBench: Evaluating the Emotional Intelligence of Large Language Models

• E-CORE: Emotion Correlation Enhanced Empathetic Dialogue Generation EMNLP 2023

• Commonsense-Aware Prompting for Controllable Empathetic Dialogue Generation ODD EMNLP 2023

• [Harnessing the Power of Large Language Models for Empathetic Response Generation: Empirical Investigations and Improvements]

4.2: External Acting / Interactions

• Hello Again! LLM-powered Personalized Agent for Long-term Dialogue using memory/persona as external sources

• SAFARI: Large Language Models as Source Planner for Personalized Knowledge-grounded Dialogues ODD EMNLP 2023 🔥🔥🔥🔥🔥 related work: Self-RAG, ToolkenGPT, RAG. dependency between different sources

• ChatCoT: Tool-augmented Chain-of-Thought Reasoning on Chat-based Large Language Models EMNLP 2023

• Towards LLM-driven Dialogue State Tracking DST EMNLP 2023 instruction-tuning

• Multi-Source Multi-Type Knowledge Exploration and Exploitation for Dialogue Generation EMNLP 2023

• PLUG-AND-PLAY POLICY PLANNER FOR LARGE LANGUAGE MODEL POWERED DIALOGUE AGENTS Arxiv 2023

• [Reinforcement Learning for Optimizing RAG for Domain Chatbots] AAAI 2024 Workshop using rl to determine whether or not to retrieve for domain chatbots

• Manual-Guided Dialogue for Flexible Conversational Agents

Database

• Are LLMs All You Need for Task-Oriented Dialogue? TOD SIGDIAL 2023 all sub tasks

Memory

• MemoChat: Tuning LLMs to Use Memos for Consistent Long-Range Open-Domain Conversation ODD Arxiv 2023[Code]

• Prompted LLMs as Chatbot Modules for Long Open-domain Conversation ODD ACL 2023

Persona/Character/Profile/Role

• PK-ICR: Persona-Knowledge Interactive Multi-Context Retrieval for Grounded Dialogue EMNLP 2023 dependency between different sources

• Large Language Models Meet Harry Potter: A Dataset for Aligning Dialogue Agents with Characters Dataset EMNLP 2023

• Building Persona Consistent Dialogue Agents with Offline Reinforcement Learning EMNLP 2023

• CharacterChat: Supporting the Creation of Fictional Characters through Conversation and Progressive Manifestation with a Chatbot

• What, When, and How to ground: Designing User Persona-Aware Conversational Agents for Engaging Dialogue ACL 2023 Industry

• Building Persona Consistent Dialogue Agents with Offline Reinforcement Learning 🔥 EMNLP 2023 offline reinforcement learning

• Partner Personas Generation for Dialogue Response Generatio NAACL 2022 reinforcement learning

Multilingual

• A Systematic Study of Performance Disparities in Multilingual Task-Oriented Dialogue Systems EMNLP 2023

• xDial-Eval: A Multilingual Open-Domain Dialogue Evaluation Benchmark EMNLP 2023

4.3: Reasoning + Acting

• TPE: Towards Better Compositional Reasoning over Conceptual Tools with Multi-persona Collaboration ODD Arxiv 2023 🔥🔥🔥🔥🔥 language agent, tool learning

• Learning Retrieval Augmentation for Personalized Dialogue Generation EMNLP 2023

• TOD-Flow: Modeling the Structure of Task-Oriented Dialogues EMNLP 2023

• Enhancing Task-oriented Dialogue Systems with Generative Post-processing Networks EMNLP 2023

• Retrieval-Generation Alignment for End-to-End Task-Oriented Dialogue System EMNLP 2023 T5 and ChatGPT as generator, related to Q-TOD, Dual-Feedback?

• ChatPLUG: Open-Domain Generative Dialogue System with Internet-Augmented Instruction Tuning for Digital Human

Others

• Bootstrapping LLM-based Task-Oriented Dialogue Agents via Self-Talk TOD data augmentation

• Multi-User Chat Assistant (MUCA): a Framework Using LLMs to Facilitate Group Conversations

What's the future? Language Agents?

Multi-modal

• SIMMC 2.0: Situated Interactive Multimodal Conversational AI

Dataset Paper

• INSCIT: Information-Seeking Conversations with Mixed-Initiative Interactions TACL 2022

• DuLeMon: Long Time No See! Open-Domain Conversation with Long-Term Persona Memory ODD ACL 2022

• FoCus: Call for Customized Conversation: Customized Conversation Grounding Persona and Knowledge ODD AAAI 2022

• KdConv: A Chinese Multi-domain Dialogue Dataset Towards Multi-turn Knowledge-driven Conversation ODD ACL 2020 [code]

Others

• AUTOREPLY: Detecting Nonsense in Dialogue Introspectively with Discriminative Replies EMNLP2023

• [Instruct Once, Chat Consistently in Multiple Rounds: An Efficient Tuning Framework for Dialogue] Tuning Method

Position and Future Directions

• Personalisation within bounds: A risk taxonomy and policy framework for the alignment of large language models with personalised feedback

• SELF-DIRECTED SYNTHETIC DIALOGUES AND REVISIONS TECHNICAL REPORT

Other Useful Resourecs

1. https://www.promptingguide.ai/papers [prompting engineering papers]
2. https://github.com/iwangjian/Paper-Reading#knowledge-grounded-dialogue

To read

• Beyond Candidates : Adaptive Dialogue Agent Utilizing Persona and Knowledge

• Are Personalized Stochastic Parrots More Dangerous? Evaluating Persona Biases in Dialogue Systems

Welcome to cite our survey paper.

@misc{wang2023survey,
      title={A Survey of the Evolution of Language Model-Based Dialogue Systems},
      author={Hongru Wang and Lingzhi Wang and Yiming Du and Liang Chen and Jingyan Zhou and Yufei Wang and Kam-Fai Wong},
      year={2023},
      eprint={2311.16789},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}