# A Survey of Graph Prompting Methods: Techniques, Applications, and Challenges

**Authors:** Xuansheng Wu, Kaixiong Zhou, Mingchen Sun, Xin Wang, Ninghao Liu  
**arXiv:** [2303.07275](https://arxiv.org/abs/2303.07275)  
**Date:** 2023-03-13  
**Summary page:** https://www.summarizepaper.com/en/paper/2303.07275v2/

## Key points

- "Pre-train, prompt, predict training" paradigm in machine learning
- Leveraging pre-trained models with prompting functions for contextual cues
- Challenges in designing effective prompts, especially in complex scenarios
- Utilizing graph data to enhance prompt design efficiency and effectiveness
- Survey on graph prompting methods by Xuansheng Wu, Kaixiong Zhou, Mingchen Sun, Xin Wang, and Ninghao Liu
- Intersection between graphs and prompt design for improving model performance across applications
- Current challenges and potential future research directions in the field

## Summary

, , , , 

In the realm of machine learning, the "pre-train, prompt, predict training" paradigm has emerged as a powerful approach to developing robust models with limited labeled data. This methodology leverages a pre-trained model in conjunction with a prompting function that applies a template to input samples, thereby imbuing them with contextual cues and reframing target tasks as part of the pre-training process. While this technique shows promise, the design of effective prompts can prove arduous and time-consuming, especially in complex scenarios. To address this challenge, researchers have turned to graph data as a solution. Graphs offer a structured framework for representing knowledge by explicitly capturing relationships between entities. By incorporating graph knowledge into prompting functions, practitioners can enhance the efficiency and effectiveness of prompt design. In their survey on graph prompting methods, authors Xuansheng Wu, Kaixiong Zhou, Mingchen Sun, Xin Wang, and Ninghao Liu explore this intersection between graphs and prompt design. The survey delves into fundamental concepts of graph prompt learning and provides an overview of existing approaches for designing graph prompting functions. By augmenting prompts with graph knowledge, researchers aim to improve model performance across various applications. Additionally, the authors highlight current challenges in this field and outline potential avenues for future research and development. Overall, this comprehensive survey serves as a bridge between the realms of graphs and prompt design within machine learning.