“论文-专利”关联视角下的新兴技术识别研究

Research on Emerging Technology Identification from the Perspective of“Paper-Patent”Correlation

[目的/意义]基于“论文-专利”关联视角,文章通过新兴技术抽取与量化研究的方式识别“从论文到专利的创新链条上对未来技术发展趋势有引领作用”的新兴术语。[方法/过程]创新性地结合了Termolator算法和GPT提示学习的新术语提取方法。该方法通过对比实验,探索了GPT提示学习在术语抽取中的应用效果,并且显著提高了术语抽取的准确性和召回率。进一步,利用Minibatch Kmeans++算法对术语识别结果进行聚类,形成技术主题,并通过多维指标量化分析方法对这些新兴技术主题进行识别和分类。[结果/结论]将新兴技术术语划分为热点型、前沿型、应用型和潜在型新兴术语,实现对技术术语主题的有效识别和分类。研究成果表明,该方法能够有效揭示大模型研究领域中对未来技术发展趋势有引领作用的新兴技术,为新兴技术术语识别提供新途径。[局限]技术术语向量化表征和新兴技术主题识别指标阈值确定存在一定局限性,需要进行进一步研究。

[Purpose/significance]Based on the perspective of“paper-patent”correlation,this paper identifies emerging terms that“lead the development trend of future technology in the innovation chain from paper to patent”through the way of emerging technology extraction and quantitative research.[Method/process]This research innovatively adopts Termolator algorithm and GPT prompt learning to extract new terms.This method explores the application effect of GPT prompt learning in term extraction through comparative experiments,and significantly improves the accuracy and recall rate of term extraction.Furthermore,Minibatch Kmeans++algorithm is used to cluster the results of term recognition,forming technical topics,and these emerging technical topics are identified and classified by multidimensional index quantitative analysis method.[Result/conclusion]This paper divides the emerging technical terms into hot,cutting-edge,applied and potential emerging terms,and realizes the effective recognition and classification of the topics of technical terms.The results show that this method can effectively reveal the emerging technologies that lead the development trend of future technologies in the field of large model research and provide a new way for the identification of emerging technology terms.[Limitations]This study has certain limitations in the vectorization representation of technical terms and the determination of the threshold of emerging technology topic recognition indicators which deserve further study.