Automated Review Generation Method Based on Large Language Models

arXiv - PHYS - Data Analysis, Statistics and Probability Pub Date : 2024-07-30 DOI:arxiv-2407.20906

Shican Wu, Xiao Ma, Dehui Luo, Lulu Li, Xiangcheng Shi, Xin Chang, Xiaoyun Lin, Ran Luo, Chunlei Pei, Zhi-Jian Zhao, Jinlong Gong

{"title":"Automated Review Generation Method Based on Large Language Models","authors":"Shican Wu, Xiao Ma, Dehui Luo, Lulu Li, Xiangcheng Shi, Xin Chang, Xiaoyun Lin, Ran Luo, Chunlei Pei, Zhi-Jian Zhao, Jinlong Gong","doi":"arxiv-2407.20906","DOIUrl":null,"url":null,"abstract":"Literature research, vital for scientific advancement, is overwhelmed by the\nvast ocean of available information. Addressing this, we propose an automated\nreview generation method based on Large Language Models (LLMs) to streamline\nliterature processing and reduce cognitive load. In case study on propane\ndehydrogenation (PDH) catalysts, our method swiftly generated comprehensive\nreviews from 343 articles, averaging seconds per article per LLM account.\nExtended analysis of 1041 articles provided deep insights into catalysts'\ncomposition, structure, and performance. Recognizing LLMs' hallucinations, we\nemployed a multi-layered quality control strategy, ensuring our method's\nreliability and effective hallucination mitigation. Expert verification\nconfirms the accuracy and citation integrity of generated reviews,\ndemonstrating LLM hallucination risks reduced to below 0.5% with over 95%\nconfidence. Released Windows application enables one-click review generation,\naiding researchers in tracking advancements and recommending literature. This\napproach showcases LLMs' role in enhancing scientific research productivity and\nsets the stage for further exploration.","PeriodicalId":501065,"journal":{"name":"arXiv - PHYS - Data Analysis, Statistics and Probability","volume":"50 1","pages":""},"PeriodicalIF":0.0000,"publicationDate":"2024-07-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"arXiv - PHYS - Data Analysis, Statistics and Probability","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/arxiv-2407.20906","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 0

Abstract

Literature research, vital for scientific advancement, is overwhelmed by the vast ocean of available information. Addressing this, we propose an automated review generation method based on Large Language Models (LLMs) to streamline literature processing and reduce cognitive load. In case study on propane dehydrogenation (PDH) catalysts, our method swiftly generated comprehensive reviews from 343 articles, averaging seconds per article per LLM account. Extended analysis of 1041 articles provided deep insights into catalysts' composition, structure, and performance. Recognizing LLMs' hallucinations, we employed a multi-layered quality control strategy, ensuring our method's reliability and effective hallucination mitigation. Expert verification confirms the accuracy and citation integrity of generated reviews, demonstrating LLM hallucination risks reduced to below 0.5% with over 95% confidence. Released Windows application enables one-click review generation, aiding researchers in tracking advancements and recommending literature. This approach showcases LLMs' role in enhancing scientific research productivity and sets the stage for further exploration.

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

基于大型语言模型的自动评论生成方法

文献研究对科学进步至关重要，但面对浩如烟海的可用信息，文献研究显得力不从心。针对这一问题，我们提出了一种基于大语言模型（LLM）的自动综述生成方法，以简化文献处理并减轻认知负荷。在关于丙烷氢化（PDH）催化剂的案例研究中，我们的方法从 343 篇文章中迅速生成了综合评论，每个 LLM 账户平均每篇文章只需几秒钟。认识到 LLM 的幻觉，我们采用了多层质量控制策略，确保我们的方法可靠并有效地减少幻觉。专家验证确认了所生成评论的准确性和引文的完整性，表明 LLM 的幻觉风险降低到 0.5% 以下，可信度超过 95%。发布的 Windows 应用程序可以一键生成评论，帮助研究人员跟踪进展和推荐文献。这种方法展示了 LLM 在提高科研生产力方面的作用，并为进一步探索奠定了基础。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文去求助

来源期刊

arXiv - PHYS - Data Analysis, Statistics and Probability

自引率

0.00%

发文量

期刊最新文献

PASS: An Asynchronous Probabilistic Processor for Next Generation Intelligence Astrometric Binary Classification Via Artificial Neural Networks XENONnT Analysis: Signal Reconstruction, Calibration and Event Selection Converting sWeights to Probabilities with Density Ratios Challenges and perspectives in recurrence analyses of event time series