Identifying FinTech Innovations with Patent Data: A Combination of Textual Analysis and Machine-Learning Techniques

Classification of FinTech Patents by Machine Learning and Deep Learning Reveals Trends of FinTech Development in China

“…Although there is no unified classification of FinTech patents, ongoing updates are gradually converging with the definition of FinTech. In this study, we referred to previous FinTech research [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] and research on the development of FinTech in China [29][30][31] that has analyzed and summarized the main characteristics of FinTech to divide the patent data into seven categories: FinTech-unrelated, payment, loan, insurance, security, data analysis, and investment. Table 3 provides the specific classifications.…”

“…e present study fills this research gap by identifying and classifying Chinese FinTech patents to analyze the technology layout and development status of FinTech in China. After reviewing previous research on the distribution of FinTech patents in the US and Europe [8][9][10], we screened patent application data from 2013 (the first year of Internet finance in China [11]) to 2020 using machine learning technology to classify the patent text. Patents related to FinTech were identified and classified according to different financial industry applications.…”

Competition and innovation in the financial sector: evidence from the rise of FinTech start-ups

“…3 Appendix Table 12 shows how our FinTech patents dataset compares to previous classifications in Chen et al (2019) and Xu et al (2020). Our taxonomy aims to capture innovations that pursue the integration of more sophisticated IT tools and data science solutions in financial products.…”

“…We also compare the results obtained from BERT models with text classifiers such as CNN (Convolutional Neural Networks) and RNN (Recurrent Neural Networks), which take BERT embedding as input. Finally, to allow comparison with previous FinTech patent classifications and Xu et al (2020), we also contrast the results of the deep learning models with traditional machine learning approaches such as: support vector machines (SVM), neural networks, as well as Naïve Bayes (NB), k-nearest neighbour (kNN), random forest, and gradient boosting. Appendix 6.2 details the machine learning methodology employed.…”

“…This section provides further details on the construction of our FinTech patent dataset. It also compares our dataset with that of Chen et al (2019) and Xu et al (2020), which are, to the best of our knowledge, the closest related studies that employ machine learning approaches to classifying FinTech innovations. A summary of the dataset characteristics and filtering steps employed is presented in Table 12.…”