Recognition of promising technologies considering inventor and assignee's historic performance: A machine learning approach

Recognition of promising technologies is important for enterprises that are eager to occupy the center position in future markets, considering it can provide valuable R&D (research and development) intelligence and industrial reform signal. Accordingly, many approaches have been introduced in the literature to identify promising technologies; however, most previous studies have relied heavily on technologies' bibliometric and text features. The promisingness of a technology is often determined by various other factors, including the inventor and assignee's historic performance (IAHP) features. To overcome the limitations of previous approaches, we propose a machine learning approach that integrates the bibliometric, text, and IAHP features to accomplish promising technology recognition (named MLIFPR) in this paper. The proposed MLIFPR approach was applied to five fields, including the electrical communication, ship, road construction, electric power, and electric vehicle fields, and its usability was verified. Experiments showed that the approach considering bibliometric, text, and IAHP features achieved an average promising technology recognition precision of 95.9 %, which outperformed existing studies. The recognition performance was significantly improved by >3.3 % after the addition of IAHP features. The proposed MLIFPR approach of this study is a novel data mining tool to assist enterprises in developing and following unhatched disruptive technologies to occupy a leading position in the future market. Besides, the introduced IAHP features as new low-dimensional signals representing the technologies' promisingness can provide a reference for other technological innovation works.