月球与行星科学多模态大模型的概念及应用初探

A first look at the concept and application of multimodal large models for lunar and planetary science

自2022年底一系列大型语言模型(大模型)发布以来,以大模型为研究热点的新一代人工智能、(AI)技术在各行各业引起高度重视,随即出现的专业大模型更将重塑千行百业。月球与行星科学领域作为科学技术前沿的代表,建立该领域的专业大模型是今后的必然趋势。本文在多模态大模型的基础上首次提出了月球与行星科学多模态大模型的概念,并从其定义、设计及应用场景等细节展开详细阐述和讨论。模型旨在帮助更多人认识月球并了解月球,同时协助更多学者处理并分析月球与行星数据,促进挖掘月球与行星大数据间的隐式或非线性关系,主动发现并解决未知科学问题,率先推动大数据驱动科学发现这一新的研究范式在行星科学领域落地生根。

Since the release of a series of Large Language Model(LLM)at the end of 2022,the new generation of Artificial Intelligence(AI)technology LLM has become a research hotspot and has attracted great attention,and the ensuing domain-specific LLM will definitely reshape thousands of industries.The field of lunar and planetary science is a representative of the frontiers of science and technology,and the establishment of domain-specific LLM in this field is an inevitable trend for the future.This article proposes the concept of Lunar and Planetary Multimodal Large Language Model(LP-MLLM)for the first time,and elaborates and discusses the details of its definition,design,and application scenarios.LP-MLLM aims to help more people know the Moon and understand it,as well as assist more scholars to process and analyze the lunar and planetary big data,promote the excavation of implicit or non-linear relationships among different lunar and planetary data,proactively discover and solve unknown scientific problems,all of which are leading the way in driving the new research paradigm of big data-driven scientific discovery to take root in lunar and planetary science.