基于专利数据的ATR技术融合关系预测研究

Research on ATR Technology Fusion Relationship Prediction Based on Patent Data

文中旨在实现ATR领域关键技术融合关系的发展趋势预测,并对基于机器学习的技术融合预测方法进行了改进,以提高预测精度。文中以全球ATR技术领域的专利数据为数据源,搭建了一种CNN与SVM相结合的深度学习模型来进行技术融合,预测未来趋势,并将CNN自动学习特征与SVM在非线性问题分类上的优势结合起来,预测到2025年ATR技术领域将出现融合概率较高的技术类别。模型在预测实验中最高能达到90%精度,较之现有研究有所提高。另外,对行业发展情况进行了描述和分析,预测结果显示,〈G06K,G06T〉,〈G06K,G06V〉,〈G01S,G06K〉等IPC对,在未来一段时间内出现技术融合的概率相对较高。

The purpose of this paper is to realize the development trend prediction of the fusion relationship of key technologies in the ATR field,and to improve the technology fusion prediction method based on machine learning to improve the prediction accuracy.In this paper,a deep learning model combining CNN and SVM is built to perform technology fusion and predict future trends.Combining the automatic learning features of CNN with the advantages of SVM in nonlinear problem classification,it is predicted that by 2025,there will be technology categories with high fusion probability in the ATR technology field.The model can achieve up to 90%accuracy in prediction experiments,which is improved compared with existing research.In addition,the development of the industry is described and analyzed,and the forecast results show that IPC pairs such as〈G06K,G06T〉,〈G06K,G06V〉,〈G01S,G06K〉have a relatively high probability of technology integration in the next period of time.