A Deep Learning-Based Early Patent Quality Recognition Model

Patent quality is important for the operation of the intellectual property market and the strategic layout of enterprises. However, the number of patent applications is increasing every year and only a small part of them are used. In this study, we propose a classification model based on deep learning to identify the quality of early patents. According to invention patents, utility model patents and design patents, the abstract, claims and technical efficiency phrases of each patent are taken as text features; take patent “reputation”, patent protection scope and patent technology diffusion as digital features simultaneously. Finally, the combination of digital and text features for each category of patents is used for early patent quality classification prediction. Theoretically, this model combines patent text features and digital features more comprehensively as the evaluation of early patent quality, and can assist patent market layout personnel to quickly screen early valuable patents.