A Quality Evaluation Method for Invalid Patents Based on Mixed Methods
Invalid patents serve as a valuable reference for small and medium-sized enterprises(SMEs)in technological research and development.Identifying and utilizing high quality invalid patents is crucial for bridging the gap in R&D resources and avoiding infringement risks.This article selects 13 indicators to evaluate the quality of invalid patents from three dimensions:technical value,economic value,and the scope of patentees'innovation activities.It calculates the comprehensive scores of invalid patents by applying the entropy-weighted TOPSIS model,and adopts the interquartile range method to set the threshold for hierarchical evaluation,so as to evaluate the quality of invalid patents from the perspective of index calculation.It employs the pre-trained Sentence-BERT model based on deep learning to vectorize the patent abstracts,and identifies the outlier patents using the density-based outlier detection algorithm,so as to evaluate the quality of invalid patents from the perspective of content analysis.The proposed method is empirically verified by taking authorized invalid patents in the field of"Artificial Intelligence"as an example,and the results are validated in terms of indicators,methods,and policies,thus providing innovative solutions for SMEs to quickly select high-quality invalid patents and bridge their technology gaps.
NETL
NSTL
万方数据
