Semantically-based patent thicket identification

Abstract

Patent thickets have been identified as a major stumbling block in the development of new technologies, creating the need to accurately identify thicket membership. Various citations-based methodologies (Graevenitz et al, 2011;Clarkson, 2005) have been proposed, which have relied on broad survey results (Cohen et al, 2000) for validation. Expert evaluation is an alternative direct method of judging thicket membership at the individual patent level. While this method potentially is robust to drafting and jurisdictional differences in patent design, it is also costly to use on a large scale. We employ a natural language processing technique, which does not carry these large costs, to proxy expert views closely. Furthermore, we investigate the relation between our semantic measure and citation based measures, finding them quite distinct. We then combine a variety of thicket indicators into a statistical model to assess the probability that a newly added patent belongs to a thicket. We also study the role each measure plays, as part of creating a prospective screening model that could improve efficiency of the patent system, in response to Lemley (2001).

Truly standard‐essential patents? A semantics‐based analysis

“…Both deGrazia et al (2020) and Kelly et al (2021) borrow the term frequency‐inverse document frequency (tf‐idf) approach of Younge and Kuhn (2016). 13 Gątkowski and Dietl (2020) calculate a latent semantic similarity model based on full‐text information from US patents to identify patent thickets. Whalen et al (2020) and Hain et al (2022) both train embedding models to measure technological similarity between patents.…”

“…Both deGrazia et al ( 2020) and Kelly et al (2021) borrow the term frequency-inverse document frequency (tf-idf) approach of Younge and Kuhn (2016). 13 Gątkowski and Dietl (2020) Despite their different approaches to measure textual similarity, all these studies use a homogeneous corpus; that is, texts within the patent universe. One notable exception in this matter is the early study by Magerman et al (2009).…”

“…To give particular words and concepts in a given text more importance, the algorithm applies weighting methods (i.e., term frequency and entropy). In line with many other approaches, such as Gątkowski and Dietl (2020), every text input is positioned in a reduced vector space, where one dimension can represent multiple words with the same semantic meaning. The cosine distance between any two vectors expresses their semantic similarity with each other.…”

“…In this study, we introduce a method to approximate standard essentiality based on the semantic similarity between patents and technical standards. Over the last decade, text‐based similarity measures have become a popular tool to assess the technological relatedness between patents (e.g., Arts et al, 2018; deGrazia et al, 2020; Gątkowski & Dietl, 2020; Hain et al, 2022; Natterer, 2016; Younge & Kuhn, 2016). 2 Extending this approach, we illustrate that natural language processing can also be used to calculate textual similarity between patents and standard documents.…”

Coopetition - collaboration between competitors

“…Coopetition partners can hedge against high competition intensity in an industry together (Ritala, 2012). This could for instance be done by forming a patent thicket -a dense web of overlapping patents that offers even more effective protection, as it impedes search and innovation efforts from other firms in the market (Gątkowski et al, 2020;von Graevenitz et al, 2011). In forming such a thicket, partners thus establish a more beneficial industry position for themselves through preventing others from introducing certain technologies and innovations.…”