基于xDeepFM模型的关键核心技术识别及竞争态势分析——以通信设备制造产业为例

Identification of Key Core Technologies and Analysis of Competitive Situation Based on xDeepFM Model:Take the ICT Equipment Manufacturing Industry as an Example

[研究目的]基于“逆变”情景,通过对通信产业的关键核心技术进行识别,以探寻逆变动力源及其竞争态势演化。[研究方法]梳理关键核心技术的定义、特征及识别方法,选取诺基亚、爱立信、华为及中兴4家通信设备企业自1984年起在全球申请的百万条专利为样本,构建基于因子分解机的深度预测神经网络模型(xDeepFM),实现关键核心技术识别与竞争态势分析。[研究结论]基于xDeepFM构建的识别模型能够有效在浩繁的专利数据中完成关键核心技术识别。通过对识别后的关键核心技术进行主题分析,评估通信产业关键核心技术竞争态势,为识别关键核心技术特征、打赢关键核心技术攻坚战提供情报支持。

[Research purpose]Based on the"inversion"scenario,this paper identifies the key core technologies of the telecommunication industry to explore the power source of the inversion and the evolution of its competitive situation.[Research method]The study sorts out the definitions,characteristics and identification methods of key core technologies,selects one million patents applied globally since 1984 by four communication equipment companies,Nokia,Ericsson,Huawei and ZTE,as a sample,constructs a deep prediction neural network model(xDeepFM)based on factorization machine,and realizes key core technology identification and competitive situation analysis.[Research conclusion]The identification model built based on xDeepFM can effectively identify and evaluate key core technologies in massive patent data.Through thematic analysis of the identified key core technologies,the competition situation analysis of key core technologies in the communications industry is described and evaluated to provide intelligence support for identifying the characteristics of key core technologies and win the battle against key core technologies.