文章信息/Info

Title:

Research Progress of Large Language Models in the Field of Materials Science

文章編號:

1674-3962(2025)05-0409-15

作者:

閆姿霓; 饒梓元; 曾小勤

1. 上海交通大學 輕合金精密成型國家研究中心，上海 200240 2. 上海交通大學 金屬基復合材料國家重點實驗室，上海 200240

Author(s):

YAN Zini; RAO Ziyuan; ZENG Xiaoqin

1.National Engineering Research Center of Light Alloy Net Forming，Shanghai Jiao Tong University, Shanghai 200240，China 2.State Key Laboratory of Metal Matrix Composites, Shanghai Jiao Tong University, Shanghai 200240, China

關鍵詞:

大語言模型; 自然語言處理; 機器學習; 人工智能; 可解釋性; 材料科學

Keywords:

large language model;  natural language processing;  machine learning;  artificial intelligence;  interpretability;  materials science

分類號:

TB30; TP18

DOI:

10.7502/j.issn.1674-3962.202409019

文獻標志碼:

A

摘要:

數據驅動方法已成為材料研究的重要趨勢，大語言模型的出現不僅為材料科學領域非結構化數據的處理提供了潛在的解決方案，還能夠通過其強大的語言理解和生成能力，助力科研中的知識發現、自動化分析、可解釋性提升以及多模塊協同操作，從而可推動材料科學研究的效率提升與創新突破。從大語言模型的理論基礎出發，探討其重要功能、優化方法及其在材料科學中的應用。特別是，大語言模型能夠有效處理和提取非結構化文本中的關鍵信息，幫助加速材料的發現與設計。最后還展望了大語言模型在人工智能與材料科學融合領域的未來發展前景，指出其在推動材料科學研究自動化和智能化方面的巨大潛力。

Abstract:

Data-driven approaches have become an important trend in materials research. The emergence of large language models not only provides a potential solution for unstructured data processing in the field of materials science, but also, through their powerful language understanding and generation capabilities, facilitates knowledge discovery, automated analysis, improved interpretability, and multi-module collaborative operations, thus enhancing research efficiency and driving breakthroughs in materials science. This article explores the theoretical foundations of large language models, examines their key functions, optimization methods, and applications in materials science. In particular, large language models can effectively process and extract key information from unstructured texts, helping to accelerate material discovery and design. This article also envisions the future development of large language models in the integration of artificial intelligence and materials science, highlighting their significant potential in advancing the automation and intelligence of materials research.