Identifying Potential Standard Essential Patents Based on Text Mining and Generative Topographic Mapping

摘要

The identification of the potential standards essential patents (SEPs) can make great contributions to not only the technology management theory, but also to the real practices of establishment and development of enterprise competitiveness and standardization strategy. However, despite the importance of identifying potential SEPs, the approaches to identify potential SEPs lack of adequate mining of existing technical standard text and validation of identification results based on standard updates. Therefore, in this paper, we contribute to resolving this issue by proposing a research model that integrates text mining and the generative topographic mapping (GTM) to effectively identify and verity the potential SEPs based on existing and updated standards. The universal terrestrial radio access (UTRA) technology is selected as a case study. In this case, the TF-IDF algorithm and the Latent Dirichlet Allocation (LDA) method are applied to analyze the keywords and technical theme of standard and patent documents, and GTM is used to construct standard and patent map, then the two maps are mapped by the improved similarity algorithm we proposed. Finally, 39 potential SEPs of the technology are identified, 24 of which have been verified, and the other 15 are likely to be included in subsequent standard version. This paper will contribute to the identification of SEPs methodology, and will be of interest to UTRA technology research and development experts.