Semantic Analysis for Assessing the Standard-Essentiality of Patents—A Proof of Concept

Abstract

Interoperability between technical devices is crucial for manufacturers as well as consumers and is largely facilitated by the development and adoption of technical standards. These standards, which are prominently utilized in information and communication technology, are developed by standard setting organizations such as the European Telecommunications Standards Institute. Companies involved in this process can influence the standards to secure competitive advantages, often by declaring their patents as standard-essential. However, the lack of a rigorous evaluation of these declarations can lead to issues such as overdeclaration or underdeclaration of standard-essential patents (SEPs). Determining the SEP status is complex, involves technical and sometimes economic considerations, and is typically a time-consuming manual process. Recent research has explored semantic and machine learning methods for automating SEP assessment, although these methods primarily provide probabilistic outcomes. Therefore, we provide a proof of concept, which aims to improve existing semantic analysis approaches by addressing the distinct linguistic and structural features of technical standards and patents. The results indicate which type of pre-processing and which text fragment from a selected technical standard and US-patents yields in the highest semantic similarity. This has methodological implications for researchers, as its sheds light on the assessment of standard-essentiality. Furthermore, managerial implications arise for companies, policy makers and patent examiners, who can use our approach for the assessment of SEPs and as yet undeclared patents.