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

Financial technology, or FinTech, has recently attracted considerable attention both in the financial industry and academia. It covers a large range of technologies, including big data, cloud computing, and cryptocurrency, and is widely used in the finance industry. Despite the broad application of FinTech, little academic research has explored the development of this new wave of technological innovations. Our study aims to identify, classify, and track the development of FinTech innovations using patent data. A difficulty is that there are no accurate International Patent Classification (IPC) codes that we can refer to as FinTech innovations. Hence, in this paper we provide a comprehensive method for identifying FinTech patents. We first use a text-based filtering technique to locate potential FinTech patents and get a data set comprising 37,156 records. We then construct a training sample of FinTech patents by reading those patent files manually. Next, textual analysis and machine-learning techniques are applied to identify all FinTech patents in the whole data set, based on the initial sample. We classify FinTech patents into seven categories according to the key underlying technologies and track the development of each category. Thus a whole picture of FinTech innovations is formed.