References & Citations
Computer Science > Digital Libraries
Title: A Text-Embedding-based Approach to Measure Patent-to-Patent Technological Similarity -- Workflow, Code, and Applications
(Submitted on 27 Mar 2020 (v1), last revised 10 Nov 2021 (this version, v4))
Abstract: This paper describes an efficiently scalable approach to measure technological similarity between patents by combining embedding techniques from natural language processing with nearest-neighbor approximation. Using this methodology we are able to compute existing similarities between all patents, which in turn enables us to represent the whole patent universe as a technological network. We validate both technological signature and similarity in various ways, and demonstrate at the case of electric vehicle technologies their usefulness to measure knowledge flows, map technological change, and create patent quality indicators. Thereby the paper contributes to the growing literature on text-based indicators for patent analysis. We provide thorough documentations of the method, including all code, indicators, and intermediate outputs at this https URL
Submission history
From: Daniel Hain PhD. [view email][v1] Fri, 27 Mar 2020 09:58:09 GMT (4847kb)
[v2] Mon, 1 Mar 2021 14:33:17 GMT (4969kb)
[v3] Mon, 28 Jun 2021 11:42:43 GMT (1280kb)
[v4] Wed, 10 Nov 2021 14:54:51 GMT (2191kb)
Link back to: arXiv, form interface, contact.