| 余辉,吴昀璟,夏文蕾,周晶.基于生成式人工智能的技术需求预见方法研究——以新能源汽车技术为例[J].情报工程,2025,11(5):096-105 |
| 基于生成式人工智能的技术需求预见方法研究——以新能源汽车技术为例 |
| Research on Technological Demands Forecasting Methods Based on Generative Artificial Intelligence: A Case Study of New Energy Vehicle Technology |
| |
| DOI: |
| 中文关键词: 生成式人工智能;技术需求;需求感知;检索增强;科技情报 |
| 英文关键词: Generative Artificial Intelligence; Technological Demand; Demand Perception; Retrieval Enhancement; Scientific and Technological Intelligence |
| 基金项目:河南省哲学社会科学规划项目“基于复杂适应性的平台型媒体协同治理研究”(2025BXW003);河南大学本科教育教学改革研究与实践项目“数智时代卓越传媒人才双创教育与专业教育融合发展研究”(HDXJJG2024-141)。 |
| 作者 | 单位 | | 余辉 | 1. 湖北工业大学经济与管理学院 武汉 430068;2. 湖北农村社会管理创新研究中心 武汉 430068; | | 吴昀璟 | 1. 湖北工业大学经济与管理学院 武汉 430068;2. 湖北农村社会管理创新研究中心 武汉 430068;3. 武汉大学信息资源研究中心 武汉 430072 | | 夏文蕾 | 武汉理工大学创业学院 武汉 430070 | | 周晶 | 1. 湖北工业大学经济与管理学院 武汉 430068;2. 湖北农村社会管理创新研究中心 武汉 430068; |
|
| 摘要点击次数: 13 |
| 全文下载次数: 14 |
| 中文摘要: |
| [目的/意义]基于生成式人工智能预见未来技术需求,为加快未来产业布局,实现产业升级,构筑竞争新优势提供参考。[方法/过程]提出基于生成式人工智能的技术需求预见方法,以新能源汽车领域为例,基于技术需求弱信号感知,选取电机技术、电池技术、安全技术和智能技术4 个技术领域进行任务设计,通过生成式人工智能大模型执行任务,运用检索增强技术优化初始模型,并结合行业专家经验评估模型生成内容的正确性和新颖性。[局限]所提出的模型在技术内容生成的合理性和逻辑性上有所欠缺。[结果/结论]研究结果显示,优化后的ChatGLM3-6B 在正确理解任务的基础上,生成4 个领域的技术内容新颖程度分别达到了0.79、0.72、0.88 和0.93。提出的技术需求预见方法,能生成新颖且合理的技术需求内容。 |
| 英文摘要: |
| [Objective/Significance] To provide references for accelerating future industrial layout, achieving industrial upgrading,and building competitive new advantages based on the foresight of future technological demands using generative artificial intelligence. [Methods/Processes] This study proposes a method for predicting technological demands based on generative artificial intelligence. Taking the field of new energy vehicles as an example, four technological domains including motor technology, battery technology, safety technology, and intelligent technology were selected based on weak signal perception of technological demands. Task design was conducted, and a generative artificial intelligence large model was employed to execute tasks. Retrieval-enhanced techniques were applied to optimize the initial model, and the correctness and novelty of the generated content were evaluated by industry experts’ experience. [Limitations] The model exhibits some shortcomings in the rationality and logical consistency of the generated technical content.[Results/Conclusions] The research results show that the optimized ChatGLM3-6B, based on the correct understanding of the task, achieved novelty scores of 0.79, 0.72, 0.88, and 0.93 in the generation of technical content across four fields. The proposed method for forecasting technological requirements is capable of generating novel and reasonable technical requirement content. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |