Large Language Models Know Your Contextual Search Intent: A Prompting Framework for Conversational Search

by   Kelong Mao, et al.
Renmin University of China

In this paper, we present a prompting framework called LLMCS that leverages large language models, such as code-davinci-002 of GPT-3, to perform few-shot conversational query rewriting for conversational search. We explore three prompting methods to generate multiple query rewrites and hypothetical responses, and propose aggregating them into an integrated representation that can robustly represent the user's real contextual search intent. Experimental results on two conversational search datasets, including CAst-19 and CAsT-20, show that our approach achieves significant improvements in search effectiveness over existing baselines and manual rewrites. Notably, LLMCS can significantly outperform the state-of-the-art baselines by up to +5.9% and +32.9% w.r.t. NDCG@3 on CAsT-19 and CAsT-20, highlighting the vast potential of large language models for conversational search. Our code will be released at


page 1

page 2

page 3

page 4


CONVERSER: Few-Shot Conversational Dense Retrieval with Synthetic Data Generation

Conversational search provides a natural interface for information retri...

ConvGQR: Generative Query Reformulation for Conversational Search

In conversational search, the user's real search intent for the current ...

Zero-shot Query Reformulation for Conversational Search

As the popularity of voice assistants continues to surge, conversational...

A Conversational Paradigm for Program Synthesis

Program synthesis strives to generate a computer program as a solution t...

How Can We Know What Language Models Know?

Recent work has presented intriguing results examining the knowledge con...

Large Language Models Need Holistically Thought in Medical Conversational QA

The medical conversational question answering (CQA) system aims at provi...

Fundamentals of Generative Large Language Models and Perspectives in Cyber-Defense

Generative Language Models gained significant attention in late 2022 / e...

Please sign up or login with your details

Forgot password? Click here to reset