Scholarlink

<p>This study proposes a novel Zero-Shot RISE Link Prediction (ZSRLP) Prompting framework, especially for performing Link Prediction (LP) tasks through Large Language Models (LLM), that integrates the benefits of Zero-Shot Prompt (ZSP) engineering and the RISE prompt framework to obtain user-satisfactory responses from LLM without any explicit fine-tuning or prior training of model. Also, a Natural Language Simplification (NLS) approach is introduced explicitly for LP-based Explainable Artificial Intelligence (XAI) results. The proposed ZSRLP has been tested on 5 popular LLMs explicitly for LP tasks and can be extended to any other domain as well. The results show that the 5 LLMs performed equivalently good, but ChatGPT, Grok AI, and DeepSeek performed better LP-based analysis.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Zero-shot rise prompting framework and natural Language simplification for trustworthy explainable link prediction

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Mridula Dwivedi,
  • Vipin Saxena,
  • Babita Pandey

摘要

This study proposes a novel Zero-Shot RISE Link Prediction (ZSRLP) Prompting framework, especially for performing Link Prediction (LP) tasks through Large Language Models (LLM), that integrates the benefits of Zero-Shot Prompt (ZSP) engineering and the RISE prompt framework to obtain user-satisfactory responses from LLM without any explicit fine-tuning or prior training of model. Also, a Natural Language Simplification (NLS) approach is introduced explicitly for LP-based Explainable Artificial Intelligence (XAI) results. The proposed ZSRLP has been tested on 5 popular LLMs explicitly for LP tasks and can be extended to any other domain as well. The results show that the 5 LLMs performed equivalently good, but ChatGPT, Grok AI, and DeepSeek performed better LP-based analysis.