基于离群专利和多维技术创新地图的技术机会识别路径研究

Path of Technology Opportunity Identification Based on Outlier Patents and Multi-dimensional Technology Innovation Map

[目的/意义]鉴于离群专利所蕴含的技术创新重要信息能够为企业高效开展创新活动提供一手数据,故依托多维技术创新地图构建基于离群专利识别技术机会的路径。[方法/过程]首先,运用BERT-LOF算法(Local Outlier Factor)提取候选离群专利,并通过新颖性、应用范围和发展能力等指标筛选出离群专利;其次,运用BERTopic主题建模技术从离群专利中提取技术主题和创新元素,并依托多维技术创新地图对创新元素进行维度划分,进而与多种创新法则耦合识别出多个具象技术机会;再次,利用直觉模糊层次分析法对具象技术创新机会进行评价;最后,以3D打印技术为例,阐述该技术机会识别路径的应用过程。[结果/结论]BERT-LOF离群检测模型不仅有效提高了离群专利的提取精度,而且研判的技术机会能够为企业提高创新效率、降低创新成本提供科学的决策参考依据。

[Purpose/significance] In view of the important information of technical innovation contained in outlier patents can provide first-hand data for enterprises to carry out innovation activities efficiently,this paper constructs a path based on outlier patent identification technology opportunities based on multi-dimensional technology innovation map.[Method/process] Firstly,the BERT-LOF algorithm(Local Outlier Factor) is used to extract candidate outlier patents,and the outlier patents are screened out by novelty,application scope and development ability.Secondly,the BERTopic topic modeling technology is used to extract the technology topics and innovation elements from the outlier patents,and the innovation elements are divided into dimensions based on the multidimensional map of technical innovation,and then coupled with a variety of innovation rules to identify a number of figurative technology opportunities.Again,using intuitionistic fuzzy analytic hierarchy process to evaluate figurative technology innovation opportunities.Finally,taking 3D printing technology as an example,the application process of opportunity recognition path is described.[Result/conclusion] The results show that the BERT-LOF outlier detection model proposed in this paper not only effectively improves the extraction accuracy of outlier patents,but also provides a scientific decision-making reference for enterprises to improve innovation efficiency and reduce innovation costs.