Nova: An Iterative Planning and Search Approach to Enhance Novelty and Diversity of LLM Generated Ideas

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Hu, Xiang, Fu, Hongyu, Wang, Jinge, Wang, Yifeng, Li, Zhikun, Xu, Renjun, Lu, Yu, Jin, Yaochu, Pan, Lili, Lan, Zhenzhong
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916455683981312
author Hu, Xiang
Fu, Hongyu
Wang, Jinge
Wang, Yifeng
Li, Zhikun
Xu, Renjun
Lu, Yu
Jin, Yaochu
Pan, Lili
Lan, Zhenzhong
author_facet Hu, Xiang
Fu, Hongyu
Wang, Jinge
Wang, Yifeng
Li, Zhikun
Xu, Renjun
Lu, Yu
Jin, Yaochu
Pan, Lili
Lan, Zhenzhong
contents Scientific innovation is pivotal for humanity, and harnessing large language models (LLMs) to generate research ideas could transform discovery. However, existing LLMs often produce simplistic and repetitive suggestions due to their limited ability in acquiring external knowledge for innovation. To address this problem, we introduce an enhanced planning and search methodology designed to boost the creative potential of LLM-based systems. Our approach involves an iterative process to purposely plan the retrieval of external knowledge, progressively enriching the idea generation with broader and deeper insights. Validation through automated and human assessments indicates that our framework substantially elevates the quality of generated ideas, particularly in novelty and diversity. The number of unique novel ideas produced by our framework is 3.4 times higher than without it. Moreover, our method outperforms the current state-of-the-art, generating at least 2.5 times more top-rated ideas based on 170 seed papers in a Swiss Tournament evaluation.
format Preprint
id arxiv_https___arxiv_org_abs_2410_14255
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Nova: An Iterative Planning and Search Approach to Enhance Novelty and Diversity of LLM Generated Ideas
Hu, Xiang
Fu, Hongyu
Wang, Jinge
Wang, Yifeng
Li, Zhikun
Xu, Renjun
Lu, Yu
Jin, Yaochu
Pan, Lili
Lan, Zhenzhong
Artificial Intelligence
Computation and Language
Scientific innovation is pivotal for humanity, and harnessing large language models (LLMs) to generate research ideas could transform discovery. However, existing LLMs often produce simplistic and repetitive suggestions due to their limited ability in acquiring external knowledge for innovation. To address this problem, we introduce an enhanced planning and search methodology designed to boost the creative potential of LLM-based systems. Our approach involves an iterative process to purposely plan the retrieval of external knowledge, progressively enriching the idea generation with broader and deeper insights. Validation through automated and human assessments indicates that our framework substantially elevates the quality of generated ideas, particularly in novelty and diversity. The number of unique novel ideas produced by our framework is 3.4 times higher than without it. Moreover, our method outperforms the current state-of-the-art, generating at least 2.5 times more top-rated ideas based on 170 seed papers in a Swiss Tournament evaluation.
title Nova: An Iterative Planning and Search Approach to Enhance Novelty and Diversity of LLM Generated Ideas
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2410.14255