| 基于多维空间专利地图及可拓学的技术创新路径识别与评价研究 |
| |
| 引用本文： | 王金凤,徐正强,冯立杰,李康.基于多维空间专利地图及可拓学的技术创新路径识别与评价研究[J].科技管理研究,2022,42(8):8-17. |
| |
| 作者姓名： | 王金凤 徐正强 冯立杰 李康 |
| |
| 作者单位： | 上海海事大学,上海 201306 |
| |
| 基金项目： | 国家科技部创新方法工作专项“中国情境下的创新方法研究与工具开发（项目编号：2018IM020300）；国家科技部创新方法工作专项“端端驱动，融合赋能”创新方法新系统研究与应用示范（项目编号：2019IM020200）；上海市科技计划项目“元易创新方法在港航物流工程与海洋装备关键技术领域的应用研究（项目编号：20040501300） |
| |
| 摘 要： | 鉴于在目前的技术机会识别中存在研判的创新路径往往较为抽象和模糊，并在很多情况下需领域专家参与解读的问题，以冷库技术为例，研究构建基于文本挖掘、机器学习算法及多维空间专利地图的技术创新路径识别模型。首先，构建技术创新路径识别框架对相关专利文献进行分词、清洗等预处理并建立知识图谱；其次，采用融合词频-逆文档频率（TF-IDF）文本挖掘方法对专利文档提取关键词，继而采用隐含狄利克雷分布（LDA）算法对主题聚类降维并萃取创新维度；再次，依据目标技术问题和目标优选创新法则耦合变换于多维空间专利地图并具象出具有现实意义、有价值前景的创新路径；最后，利用可拓学计算各创新路径综合关联度评级优选。以期减少创新成本、提高创新效率，为企业精准开展技术创新、不断提升核心竞争力提供决策参考。
|
| 关 键 词： | 创新路径识别 多维空间专利地图 可拓学 LDA TF-IDF |
| 收稿时间： | 2021/10/8 0:00:00 |
| 修稿时间： | 2022/5/8 0:00:00 |
Research on Identification and Evaluation of Technological Innovation Path Based on Multi-dimensional Space Patent Map and Extenics |
| |
| Wang Jinfeng,Xu Zhengqiang,Feng Lijie,Li Kang.Research on Identification and Evaluation of Technological Innovation Path Based on Multi-dimensional Space Patent Map and Extenics[J].Science and Technology Management Research,2022,42(8):8-17. |
| |
| Authors: | Wang Jinfeng Xu Zhengqiang Feng Lijie Li Kang |
| |
| Abstract: | Since the innovation paths in the current technology opportunity identification are often abstract and vague, and in many cases require the participation of experts in the field of interpretation, the study builds a technology innovation path identification model based on text mining, machine learning algorithms, and multidimensional spatial patent maps, taking cold storage technology as an example. The technological innovation path identification method is proposed through, and multi-dimensional spatial patent map. Firstly, the technological innovation path identification framework is constructed to pre-process the relevant patent documents by word separation and cleaning and establish the knowledge map; secondly, the fused word frequency-inverse document frequency (TF-IDF) text mining method is used to extract keywords from patent documents, followed by the implicit Dirichlet distribution (LDA) algorithm to reduce the dimensionality of topic clustering and extract innovation dimensions; again, based on the target technical problem and the target Finally, the topology is used to calculate the comprehensive correlation degree of each innovation path. To reduce innovation cost, improve innovation efficiency, and provide decision-making references for enterprises to accurately carry out technological innovation and continuously improve their core competitiveness. |
| |
| Keywords: | innovation path recognition multi-dimensional spatial patent map extenics LDA TF-IDF |
| 本文献已被 万方数据 等数据库收录！ |
| 点击此处可从《科技管理研究》浏览原始摘要信息 |
| 点击此处可从《科技管理研究》下载免费的PDF全文 |
|
