新兴技术的多指标量化识别研究--基于向量表征方法的探索

Research on Multi-Index Quantitative Recognition of Emerging Technologies:Exploration Based on Vector Representation Method

[目的/意义]立足计量视角,通过对新兴技术特征的量化评价识别“目前处于科学研究阶段、尚未完全进入产业研发落地”的新兴技术。[方法/过程]借助Node2Vec网络表征方法,从术语共现网络中学习技术术语的向量表示;以此为基础量化新兴技术“过去、现在及未来”三大时间维度特征-“融合性、新颖性及潜在的科学影响力”,用特征值筛选技术主题是否具有新兴性,由此探索得到向量表征视角下的新兴技术识别模型。最后以航空领域为例进行实证研究,验证该方法的科学性和合理性。[结果/结论]通过引入“术语向量表征”的计算视角,有效编码了术语实体间显性和隐性的关联关系,提升了新兴技术特征计算的客观性;同时结合技术的历史、当前和预测信息,从网络结构和语义特征两方面进行识别,取得了较好的效果。

[Purpose/significance]Based on a metrological perspective,this study intends to identify emerging research technologies that are"currently in the scientific research stage and have not yet fully entered the industrial R&D landing"through the quantitative evaluation of the characteristics of emerging technologies.[Method/process]Specifically,with the help of the Node2Vec network representation method,this paper learned the vector representation of technical terms from the term co-occurrence network,used this as a basis to quantify the three time-dimension characteristics of emerging technologies"past,present and future"——"fusion,novelty and potential scientific influence",used eigenvalues to filter whether technical topics were new or not,and built the emerging technology recognition model from the perspective of vector representation.Finally,taking the aviation field as an empirical study was to verify the scientific nature and rationality of the method.[Result/conclusion]Through the introduction of the computational perspective of"term vector representation",this paper effectively encoded the explicit and implicit associations between term entities,and improved the objectivity of feature calculation of emerging technologies;at the same time,combining historical,current and predicted technical information,this paper identifies them from two aspects of network structure and semantic features,and has been achieved good results.